Skip to main content

Full text of "Time Bases ( Their Design & Development )"

See other formats


TIME BASES 

(scanning generators) 



TIME BASES 

(SCANNING GENERATORS) 
Their Design and Development 

WITH NOTES ON THE CATHODE 

RAY TUBE 



By 
O. S. PUCKLE 

M.B.E., M.I.E.E. 

E.M.J. Engineering Development Ltd., Hayes, Middlesex 

Formerly Research Department, Messrs. A. C. Cossor, Ltd., London 

Felloiu and Chairman of Council of the Television Society 

With a Foreword by 
E. B. MOULLIN 

M.A., ScD., M.I.E.E. 



SECOND EDITION 
COMPLETELY REVISED 




LONDON 

CHAPMAN & HALL LTD 

37 ESSEX STREET W.C.2 
^S 1 



First published 1943 

Second Impression 1943 

Third Impression 1 944 

Fourth Impression 1945 

Fifth Impression 1947 

Second Fdition, revised, 1951 



To 
O. L. P., M. E. P. 

AND 

L. H. B. 



Bases de Temps (French translation) 1948 
Tijdbasis SchakeKngen (Dutch translation) 1949 



Catalogue No. 158 4 



PRINTED IN GREAT BRITAIN BY JARROLD AND SONS, LTD,, NORWICH 



FOREWORD 

At the end of the last century the cathode ray oscillograph was used, 
almost in its present structural form, by J. J. Thomson for measuring 

a 

the ratio — • In the last ten years or so it has become an invaluable tool 
m 

of electrotechnics and has made possible television and several other 
spectacular developments. Its use as a laboratory tool for delineation 
of wave-forms has contributed enormously in the development of 
circuits which are essential in those applications in which it is the final 
indicating device. In all such applications it is necessary to move the 
fluorescent spot rhythmically across the screen, and the devices for 
doing this are called Time Bases. They correspond very closely with 
the "indicator rig" of the indicators used in testing reciprocating 
engines and contain the same problems of attaining faithfulness of 
response; but very many forms of response are called for which have 
not been needed in testing engines. Ill-designed indicator rigs were 
often responsible for very fallacious results; high-speed engines soon 
made indicators a dangerous tool in the hands of untaught and 
unthinking users. The same dangers have been latent in the delight- 
fully versatile oscilloscopes which the electro technician has recently 
been able to purchase at a very moderate cost and to use with seductive 
case. 

Therefore an authoritative book on the construction, testing and uses 
of Time Bases is already overdue. Such a book has now been written 
by Mr. Puckle, who has made outstanding contributions to their 
development. The book is comprehensive, well written and uniformly 
of that high standard which our learned profession has long learned to 
associate with all Mr. Puckle's work. It by no means always happens 
that books are written by people who understand what they are writing 
about or who write with that care and precision which is essential 
if the reader is to be able to understand what is written. It is most 
fortunate this book has been written by Mr. Puckle, both because 
it can be understood and because he writes with wide knowledge 
which is first hand. It is to be hoped many people will read and 
use it. 

The excellent historical development of the subject gives coherence to 
the book, inspires confidence in its completeness and adds appreciably 
to its interest. 

The appendix on the behaviour of gas-discharge triodes is particu- 
larly welcome, and can do much to dispel the cloudy misapprehensions 



Vlll 



TIME BASES 



which are current in respect of their frequency limitations. Now this 
excellent little book is available many incidental difficulties which are 
apt to arise in experimental work can readily be corrected; it should 
earn the gratitude of many workers in diverse fields. 

E. B. MOULLIN 

Engineering Laboratory, Oxford 

March 1943 



PREFACE TO SECOND EDITION 

The very kind reception which has been given to the first edition of 
this book, together with the lifting of the ban on the publication of 
many war-time developments, has made it a necessary but enjoyable 
task to write a second edition. 

The most significant war-time developments in the time base field 
have been the great improvements in linearity and the greatly increased 
trace velocities which have resulted from the development of the 
Miller-capacitance and other negative feed-back types of circuit. 

In addition, the accuracy of initiation of the trace has been greatly 
increased so that the best time bases can now be considered as very 
highly accurate measuring instruments. As a result of these improve- 
ments, enormously increased accuracy in anti-aircraft fire against targets 
whose speed has now increased to more than twice the pre-war figure 
is obtainable. In the field of frequency dividers, counting circuits and 
blocking oscillators, great advances have also been made. The possibility 
of measuring times of the order of one microsecond or less with consider- 
able accuracy and rather longer times with accuracies of about one part 
in 50,000 or better, will find applications which will increase the sum of 
human knowledge and provide new tools and appliances. 

These developments from the early work of Blumlein and Newsam 
have been mainly due to Williams, Moody, Benjamin, Fleming- 
Williams, and Whiteley, but many others have made valuable contribu- 
tions. Further notable developments would almost certainly have come 
from Blumlein were it not for his tragic and untimely end in an aeroplane 
disaster while engaged on research work for the R.A.F. Unfortunately, 
all these developments have made the work of splitting the book into 
distinct subjects, under separate chapter headings, even more difficult 
than was the case when writing the first edition. There are many time 
bases which are eligible for inclusion in two, or even three, chapters. 
For this reason, the author can only repeat that he has done his best to 
ensure that a logical sequence with sufficient cross-reference shall form 
the basis of the work. 

In the present edition, Chapter IX, which deals with the subject 
of Miller-capacitance time bases is entirely new, and many modifications 
and additions have been incorporated in most of the other sections of 
the book. 

The author wishes to acknowledge the kindness of the following 
persons who have made suggestions or provided information: Sir Robert 
Watson Watt, Dr. C. Singer, Prof. F. C. Williams, Messrs. L. H. 



TIME BASES 



Bain bridge-Bell, A. V. Bedford, L. H. Bedford, G. Builder, E. Cattanes, 
G. Chavaris, A. C. Clarke, R. H. Colborne, C. L. Faudell, B. C. 
Fleming- Williams, E. Gleave, R. W. Hallows, H. den Hartog, S. Hol- 
brook, M. L. Jofeh, K. C. Johnson, A. W. Keen, P. Lebail, H. M. 
Lewis, G. L. Light, H. Moss, S. O. Pearson, R. Puleston, G. S. P. 
Scantlebury, T. B, Tomlinson and A. F. Wilkins. Mr. J. E. Wright, 
Librarian of the Institution of Electrical Engineers, has provided assis- 
tance regarding the sources of information, while the British Museum 
and Science Museum Libraries, Miss Moon, librarian of Sobell Indus- 
tries Ltd., and Mr. S. J. Preston, Dr. J. E. Best and Miss Criticos of 
E.M.L Research Ltd. have lent many books, patent specifications and 
periodicals. The following bodies have also provided information which 
is hereby gratefully acknowledged: A. C. Cossor Ltd., the Edison Swan 
Electric Co. Ltd., Electrical and Musical Industries Ltd., Ministry of 
Supply, and Mullard Radio Valve Co. Ltd. In addition, the author 
wishes once more to thank all those authors upon whose published work 
he has freely drawn and, in particular, to pay a very special tribute to 
Prof. F. C. Williams for his most excellent paper "Introduction to 
Circuit Technique for Radiolocation," and to Prof. Williams and his 
colleagues for their papers on "Linear Time Base Generators," "Elec- 
tronic Servo-simulators" and "Automatic Strobes and Recurrence- 
frequency Selectors," which formed part of the I.E.E. Radiolocation 
Convention in 1946. These papers are so excellent that the present 
author has decided to quote verbatim from them and has received 
Prof. Williams's permission to do so. The latter 's work on circuit 
analysis and synthesis is of very great value. 

The author wishes to acknowledge his grateful thanks to his wife, 
who has again typed the whole of the manuscript and without whose 
patience and perseverance neither the original book nor the second 
edition would have appeared. 

In conclusion, the author's thanks are due and very willingly 
given to Messrs. Chapman and Hall, Ltd., and, in particular, to Messrs. 
J. L. Bale, G. Parr and E, W. Hamilton for their efforts in publishing 
many reprints and the second edition under very adverse conditions. 



O. S. Puckle 



Beaconsfield, 1950 



PREFACE TO FIRST EDITION 

This book is based on a paper of the same title which the author read 
before the Institution of Electrical Engineers in February 1942. It is 
intended to cover all the more important electronic devices which are 
available for producing the time axis in television receivers, cathode 
ray oscillographs, engine indicators and similar apparatus involving 
precise timing or measurement of time intervals. 

The time base is daily being put to new uses, and it is likely that it 
will become of greater importance in the future. However, some of the 
most useful devices in this field were invented more than twenty years 
ago although they were almost unknown to the majority of electronic 
engineers until the cathode ray tube reached such a state of develop- 
ment that it presented opportunities for their exploitation. Within a 
few years, interest in time bases, trigger circuits and relaxation oscil- 
lators has grown very quickly, and inventors have produced so many 
new circuits and modifications of old ones that the historical background 
has been obscured. 

It is quite possible that the circuit arrangements which are popular 
to-day may become obsolete almost overnight, but the fundamental 
principles of the technique are well established and the student 
who masters them will have no difficulty in the assimilation of new 
developments. 

With this in mind, the author has attempted to provide an introduc- 
tion to the principles, rather than a compendium of inventions, and has 
included a brief statement of the historical aspect. Long-forgotten 
devices which have never been successfully applied, often become 
extremely valuable when improvements in other fields make their 
realization practicable. Every engineer should therefore make a point 
of learning something of what has been attempted in the past. 

In order to obtain a knowledge of time bases and to apply that know- 
ledge to a particular purpose or to develop new methods, a study of the 
effects of time constants and other valve and circuit parameters is 
essential. These subjects are discussed in the book, and a short account 
of the cathode ray tube and its employment in connection with time 
bases has been included. These sections cannot justifiably find a place in 
the main part of the book, and they have, therefore, been relegated to a 
series of appendices. In the appendix on cathode ray tubes, only the 
sealed-off low voltage cathode ray tube is considered. 

It should be noted that the term "steady" has been employed to 
denote a uni- directional condition such as is generally meant when the 



Xll 



TIME BASES 



term "D.C." is employed. This has been done because the author 
considers that such terms as D.C. potential and D.C. current are 
inadmissible. 

The division of the book into "water-tight" chapters has proved 
extremely difficult, due to the essential interrelation of many of the 
ideas which it is the aim of the book to discuss. It can only be said 
that the author has carried out this subdivision in the manner which 
has appeared to him to provide the clearest exposition. 

The author wishes to thank Dr. E. B. Mouilin, whose very kind 
interest and encouragement have led to the appearance of the paper 
mentioned above and of this book. Thanks are also due to the Institu- 
tion of Electrical Engineers for permission to use the paper as a basis 
for the book, and to Messrs. A. C. Cossor, Ltd., for permission to 
write it. 

The author would like to thank all those upon whose work he has 
freely drawn either by way of description or of reference. Assistance 
has been willingly given by many friends and colleagues, notably the 
late Mr. A D. Blumlein, Messrs. M. K. Taylor and M. L. Jofeh, Mr. 
B. C. Fleming- Williams who read the proof of Appendix I, and Messrs. 
G. Parr and T. D. Humphreys who have each supplied many useful 
references. Mr. J. Corthesy and the Institution of Electrical Engineers 
have provided details regarding the sources of information, and Mr. 
L. B. Shuffery, the Science Museum Library, and Messrs. A. C. 
Cossor, Ltd., have lent innumerable books and periodicals. He also 
wishes to thank the Marconi -E.M.I. Company for permission to 
print Fig. 2, and Mr. J. L. Bale, of Messrs. Chapman & Hall, Ltd., 
for his unfailing courtesy and assistance in the task of printing and 
publishing, the difficulties of which have been greatly increased by 
the War. 

Finally, he wishes to record his most grateful thanks to Mr. L. H. 
Bedford, a friend of many years' standing, and one who is always willing 
to help in solving a problem, and to Mr. J. W. Whitcley who has very 
effectively criticized the proofs and made many invaluable suggestions. 

If any readers find errors, or can suggest modifications which 
would improve a possible future edition, and would care to write to 
the author, c/o Messrs. Chapman & Hall, Ltd., he would be 
extremely grateful. 

O. S. Pi.'CKLE 
Edgwarf, March 1043 



SUPPLY 



A D.C. SUPPLY 



CONTENTS 

FOREWORD 

PREFACE TO SECOND EDITION 

PREFACE TO FIRST EDITION 

CHAPTER 

I. INTRODUCTION 

II. TIME BASE WAVE FORMS 

III. TYPES OF TIME BASE 

TIME BASES OPERATED FROM AN A.C. 

Sinusoidal Time Bases 

Circular Time Base 
TIME BASES OPERATED FROM 

Capacitive Time Bases 

Inductive Time Bases 
SOFT VALVE TYPES OF LINEAR TIME BASE 

Capacitive Time Bases 

Blumlein's Thyratron Time Base 

Faudell and White's Thyratron Time Base 
HARD VALVE TYPES OF LINEAR TIME BASE 
Self-operated Time Bases of the Multivibrator Type 

Abraham and Bloch's Multivibrator 

Friihauf's Time Base 

Puckle's Time Base 

Keen's Time Base 

Den Hartog and Muller's Time Base 

Moss's Time Base 

Potter's Time Base 
INVESTIGATION INTO THE PERFORMANCE OF 
A TIME BASE 
IV. TRIGGERED, SINGLE-STROKE OR EXTERNALLY OPER- 
ATED TIME BASES 

Smith's Triggered Time Base 

V. TRIGGER CIRCUITS 

The Transitron Relaxation Oscillator 
The Transitron Sinusoidal Oscillator 
The Transitron Saw-tooth Oscillator 
The Screen -coupled Transitron Flip-flop 
The Cathode-coupled Transitron Flip-flop 
TWO-VALVE TRIGGER CIRCUITS 
The Two-valve Flip-flop 
The Eccles-Jordan Trigger Circuit 



PAGE 

vii 



I 

6 

11 
1 1 

1 1 
11 

13 
14 
15 

IS 
15 
25 
28 

28 
28 
28 
36 
37 
4i 
43 
44 
47 

SO 

52 
53 

55 
56 
58 
60 

64 
70 

72 
72 
76 



xiv TIME BASES 

CHAPTER 

TWO- VALVE TRIGGER CIRCUITS— contd. 
Evans' Trigger Circuit 
Schmitt's Trigger Circuit 
Flip-flop Form of Schmitt's Trigger Circuit 
Chance's Trigger Circuit 

SECONDARY EMISSION TRIGGER CIRCUIT 
Black's Time Base 

VI. BLOCKING OSCILLATORS AND INDUCTIVE TIME BASES 
BLOCKING OSCILLATORS 

Appleton, Herd and Watson Watt's Time Base 

Kobayashi's Time Base 

Faudell and White's Time Base 

Holmes, Carlson and Tolson's Time Base 
INDUCTIVE TIME BASES 

Blumlein's Electromagnetic Time Base 

White and Blumlein's Time Base 

VII. POLAR CO-ORDINATE, MULTIPLE AND VELOCITY- 
MODULATION TIME BASES 

POLAR CO-ORDINATE TIME BASES 

Lilienfeld's Circular Time Base 

Dye's Circular Time Base 

Watson Watt, Herd and Bainbridge -Bell's Circular 
Time Base 

The Polar Co-ordinate Oscillograph 
MULTIPLE TIME BASES 

The Spiral Time Base 

Godcck's Radial Time Base 
VELOCITY-MODULATION TIME BASE 
VIII. LINEARIZATION OF THE TRACE 

The Meaning of Linearity 

Linearization of the Trace 

LINEARIZATION BY MEANS OF INVERSE 
CURVATURE 

Bedford and Stevens' Method 
Bedford's Method 
Jenkins' Method 

LINEARIZATION BY MEANS OF AN AUXILIARY 
TIME CONSTANT 
Hawkins' Method 
Tomlinson's Method 
Keen's Method 

LINEARIZATION BY MEANS OF A VARIABLE 
INDUCTANCE 

Wilson's Method 



I 'AGE 
80 

Rr 

83 
84 

85 
85 

89 

89 
89 
9i 
92 
93 

95 
95 
97 

100 

100 
100 
101 

103 
104 
105 
105 
106 

108 
112 

112 

r 14 

114 
ri4 

US 
116 

117 
117 
1 19 
122 

124 

125 



CONTENTS 



CHAPTER 



LINEARIZATION BY 
INDUCTANCE— contd 
Taylor's Method 

LINEARIZATION BY 
ARRANGEMENTS 
The Linearization 



MEANS OF A VARIABLE 



MEANS OF FEED-BACK 

of Electrostatic Deflection 
Potentials by Means of Negative Feed-back 
Norton and Groves* Method 
Newsam's Method 
The Bootstrap Circuit 
The Linearization ok Electromagnetic Deflection 
Potentials bv Means of Negative Feed-back 
Blumlein's Method 
Faudell's Method 
The Linearization of Electrostatic Deflection 
Potentials by Means of Positive Feed-back 
Beale and Stansneld's Integrating Amplifier 

METHODS OF MEASURING THE LINEARITY' AND 
VELOCITY OF A TIME BASE TRACE 

The Use of a Calibrating Oscillator for Measuring the 

Linearity of the Trace 
Laws' Pulsed Calibrating Oscillator 
The Trace Shift Method of Determining Linearity 
Method of Determining the Linearity by Measuring 

the Slope of the Flux Variations 
Methods of Measuring the Linearity of a Raster 
Method of Measuring the Characteristics of a Current 

Regulating Circuit 

IX. MILLER-CAPACITANCE TIME BASES 
General Theory 
A Practical Circuit 
Reversed Operation 
External Operation 
Attree's Time Base 
MacKay's Time Base 

THE TRANSITRON MILLER TIME BASE 
(PHANTASTRON) 

The Cathode -coupled Phantastron 
The Screen-coupled Phantastron 

TWO-VALVE MILLER TIME BASES 
The Sanatron 

The Suppressor Grid-controlled Sanatron 
The Screen Grid-controlled Sanatron 
The Cathode-controlled Sanatron 
The Grid -controlled Sanatron 



xv 



126 

128 

129 
129 
131 
i35 

i37 
137 
140 

141 
141 

142 

H3 

i47 
149 

150 
i53 

156 

158 

158 
166 
169 
169 
171 
171 

172 

173 
178 

181 
181 
181 

186 
187 
187 



XVI 



TIME BASES 



CHAPTER 

X. PUSH-PULL DEFLECTION 



PACE 
I90 



THE USE OF AMPLIFIERS FOR PRODUCING 

PUSH-PULL DEFLECTION 192 

A Single-valve Phase-reversing Amplifier 192 

Amplifiers Employing Negative Feed-back 196 

A Single-valve Push-pull Amplifier Employing Nega- 
tive Current Feed-back 196 

Single -valve Amplifiers Employing Negative Voltage 

Feed-back 198 

A Single-valve Balanced Amplifier Employing Voltage 

Feed-back 1 99 

A Single-valve Unbalanced Amplifier Employing 

Voltage Feed-back 201 

The Design of a Feed-back Amplifier Circuit for use in 

Wave -form Shaping Circuits 203 

The Concept of the Virtual Earth 204 

The Estimation of the Errors introduced by using 

Equation (3) 205 

The Output Impedance of Feed-back Amplifiers 206 

The Input Impedance of Feed-back Amplifiers 207 

The Shaping Effect of Typical Amplifier Circuits 207 

Two- valve Push-pull Amplifiers 210 

Standard Forms of Push-pull Amplifier 210 

Cathode-coupled Push-pull Amplifiers 212 

Cathode-coupled Amplifiers 213 

The Amplification of Balanced Potentials 218 

Cathode-coupled Amplifier Stages connected in Series 219 
A Cathode -coupled Amplifier Time Base Providing 

Facilities for Sweep Expansion 221 

Anode-coupled Push-pull Amplifiers 223 
An Amplifier combining the Principles of Cathode 

and Anode Coupling 226 

PROVISION OF PUSH-PULL DEFLECTION WITH- 
OUT THE USE OF AMPLIFIERS 228 

Split Time Bases 228 

I ) (Kid's Time Base 229 



Two Equivalent Time Bases Operating in Phase 
Opposition 

XI. THE SYNCHRONIZATION OF TIME BASES 
THE GENERAL PROBLEM 

Synchronization of a Soft Valve Time Base 
Synchronization of a Soft Valve Frame Time Base in 

a Television Receiver 
Synchronization of a Hard Valve Time Base 
Synchronizing -pulse Generators 



230 

232 
232 
235 

238 

239 
240 



CONTENTS 



XVll 



CHAPTER VAltt. 

XU. THE USE OF A TIME BASE FOR FREQUENCY DIVISION 

AND FOR COUNTING 242 

FREQUENCY DIVISION 242 

Frequency Division from a Sinusoid 243 

Builder's Frequency Divider 244 

Frequency Division from a Pulse 248 

Clayden's Frequency Divider 248 

The Cathode-controlled Sanatron Frequency Divider 250 
The Suppressor Grid -controlled Santrig Frequency 

Divider 25 1 

Scalariform Frequency Divider or Time Base 254 

COUNTING 256 

Benjamin's Blocking Oscillator Counter 257 

APPENDICES 259 

(For Contents, see page six) 

BIBLIOGRAPHY 377 



INDEX 



379 



APPENDICES 



I. THE CATHODE RAY TUBE 259 

The Production of a Double Beam 261 
The Advantages of a Cathode Ray Tube as an Indi- 
cating Mechanism 262 

OPERATING POTENTIALS FOR THE CATHODE 

RAY TUBE 263 

DEFLECTION METHODS 268 

DEFLECTION SENSITIVITY 269 

Electrostatic Deflection Sensitivity 269 

Electromagnetic Deflection Sensitivity 272 

BRIGHTNESS MODULATION 273 

Brightness Modulation of the Trace 273 

Unwanted Brightness Modulation 274 

BEAM VELOCITY 27s 

DISTORTION 275 

The Use of Incorrect Potentials 276 

Asymmetric Distortion 276 

Deflection Defocusing 276 

Trapezium Distortion 276 

Astigmatism 278 
The Measurement of the Distortion in a Cathode Ray 

Tube 279 

THE APPLICATION OF SHIFT POTENTIALS 285 

THE USE OF ALTERNATING POTENTIALS FOR 

THE OPERATION OF A CATHODE RAY TUBE 288 

THE USE OF THE CATHODE RAY TUBE 289 

Cross-talk 291 

The Use of a Cathode-follower Probe 293 

FLUORESCENT SCREENS 294 

Secondary Emission and Developers 294 

THE PHOTOGRAPHY OF CATHODE RAY TUBE 

TRACES 298 

Photographic Emulsions and Developers 302 

The Use of a Graticule 3°2 

Errors in Photography 303 

II. THE CURVATURE OF THE CHARGING CHARACTERISTIC 3°5 

xix 



XX 
APPENDIX 



TIME RASES 



III. 



IV, 



THE CHARACTERISTICS OF A GAS-DISCHARGE TRIODE 

USED AS A TIME BASE DISCHARGE VALVE 309 

The Gas -discharge Triode 3°9 

The Static Characteristics 3 10 

The Mechanism of a Gas-discharge Time Base 313 

DIFFERENTIATING AND INTEGRATING NETWORKS 323 

Simple Forms of Differentiating and Integrating Net- 
work 3 2 4 
High Impedance Networks 3 2 5 
Low Impedance Networks 327 
Response of Common Types of Simple Differentiating 
and Integrating Networks to Sine Waves and to Step 
Functions of Amplitude and Velocity 3 2 § 
Aids to Rapid Determination of the Shaping Effect of 

a Network 332 
Derivation of Compensating Networks 334 
Use of these Methods in Circuits containing Non- 
linear Elements 335 
Phase Shift and Attenuation in a Simple Differentiating 

Network having a Sinusoidal Input 335 
Phase Shift and Attenuation in a Simple Integrating 

Network having a Sinusoidal Input 337 

Whiteley's Differentiating Circuit 339 

The Miller -capacitance Integrator 340 

Wave -front Distortion 343 



V. SIMPLE ELECTRICAL COMPUTING NETWORKS 345 
Factors Determining the Choice of Computation 

Methods in Simulators 345 

Summary of Electrical Computation Methods 346 

VI. THE GENERATION OF SQUARE WAVES AND PULSES 351 

Cathode -coupled Transitron Square -wave Generator 355 

White's Multivibrator Pulse Generator 35^ 

The Multiar 35° 

The Inductive Pulse Generator 360 

Blocking Oscillator Pulse Generator 362 
The Use of a Cathode Ray Tube for Generating 

Wave-forms 3^3 

The Use of a Line for the Formation of Brief Pulses 364 

The Use of a Line to Delay a Signal 368 

VII. A METHOD OF CHANGING THE PHASE OF A SINUSOID 369 

VIII. THE EFFECT OF LARGE VALUES OF SHUNT CAPACI- 
TANCE IN THE ANODE CIRCUIT OF A VALVE 371 
The Use of a Multi-stage Integrating Network 372 



LIST OF TABLES 



PAGE 
10 



L Time Bases for use with Cathode Ray Tubes 

II. Figures relating to Jenkins' Method of Linearization . 118 

III. Figures explaining the Operation of a Miller Time Base 161 

IV. Networks for use in Feed-back Amplifiers 
V. Networks and their Transfer Operators 

VI. Time Bases for use without a Cathode Ray Tube 

VII. Types of Cathode Ray Tube Screen 

VIII. Some Phosphors used for Cathode Ray Tube Screens . 

IX. Emission Peaks of Cathode Ray Tube Screens and 
Sensitivity Peaks of Emulsions .... 

X. Relative Speeds of Recording Materials 
XL Relative Speeds and Fogs for 5G91 Developed in various 

Developers at 65" F. 
XII. Exposure Times for the Repeating Traces of Television 
Images ....-■»• 

XIII. Maximum Writing-speeds for Recording Transient 

Traces .....•■- 

XIV. Table of Exponents 

XV. Extinction Current in a Gas-discharge Time Base 

XVI. Response of Common Differentiating and Integrating 

Circuits to Step Functions of Amplitude and Velocity 332 

XVII. Result of the Differentiation and Integration of various 

Wave-forms .....■• 335 
XVIII. Summary of Electrical Computing Methods (Passive 

Networks and Feed-back Circuits) . . . 348 



208 
211 

242 

295 
297 

299 
299 

300 

300 

301 
306 

3i5 



CHAPTER I 
INTRODUCTION 

A time base* (or, more correctly, a time base generator) is a means of 
measuring time, the term being a modern one, usually, though not 
always, limited to an electrical method in which a potential is produced 
which varies at a known rate with respect to time. However, although 
the term "time base" is modern, instruments for measuring time have 
existed from remote antiquity. Probably the earliest form was the 
sun dial;f but the best known early instrument which was capable of 
day and night operation was the water-clock or Clepsydra used by the 
Greeks, Romans, Babylonians and Egyptians. This was a very simple 
affair, consisting of a graduated measuring jar into, or out of, which 
water was allowed to drip at a predetermined rate. If the jar of water 
is replaced by a condenser or electrical "jar," and electricity is allowed 
to "drip in or out," it becomes apparent that, fundamentally, our most 
modern method of measuring time is very like that used 2,000 or 3,000 

years ago. 

In 145 B.C. a tremendous improvement upon the Clepsydra was 
made by Ctesibius of Alexandria, who added a float and gearing so 
arranged that a hand rotated in front of a dial and pointed to the hours. 
Hour-glasses, clocks, watches and then falling mirrors and photographic 
plates followed in due course until, in the early part of the present 
century, the time base as we know it to-day made its appearance. 

During the course of the evolution of apparatus for the measurement 
of time, each advance in design has made possible higher standards of 
precision and has set a lower limit to the shortest time which can be 
measured. Whereas, in antiquity, only fractions of a day could be 
measured, we can now measure fractions of a microsecond. The 
combination of a time base with a suitable delineating instrument, such 
as a cathode ray tube (see Appendix I), provides a means of obtaining 
a graphical representation of a series of events. The ability to obtain 
this complete picture is far more valuable than the simple measurement 

* A time base, together with certain other circuits described herein in which a 
sudden change from a stable regime to a non-stable one or from one stable condition 
to another is also known as a relaxation oscillator because a state of strain is sua- 

C f The sun dial was introduced into Greece from Babylon by Animaxander (61 1 to 
S47B.C.),apupilof Thales of Miletus. It consisted, in essence, of a gnomon, a fixed up- 
right rod the direction and length of the shadow of which can be measured hour by hour. 
Records of the lengths of these shadows make it possible to determine the movements 
of the sun and to fix the dates of the two solstices (the shortest and longest days) and 
of the equinoxes (the two annual occasions when day and night arc of equal length). 
Vide Charles Singer: A Short History of Science, Clarendon Press, Oxford. 



TIME BASES 



of short time intervals since it avoids the delay involved in making a 
large number of measurements and plotting the results. The modem 
technique of time measurement may, therefore, be said to differ from 
the old, both in the order of dimensions and in the quality of the results. 
These improvements in quality are more difficult to foresee than 
changes in quantity. The clock-makers of antiquity may have guessed 
that it would eventually become possible to work with exceedingly 
short time intervals, but it is unlikely that they foresaw the connection 
between the measurement of time intervals and vision at a distance or 
the multitude of other applications of the cathode ray tube which have 
been realized during the last twenty years. 

A cathode ray tube may be put to work by itself, that is to say, 
without being associated with any kind of time base, and there are a 
number of interesting and useful applications which fall into this 
category, but their importance is small compared with the devices which 
combine a time base and a cathode ray tube. The combination is known 
as a Cathode Ray Oscillograph (C.R.O.) or Oscilloscope,* while the 
pattern seen on the screen is an oscillogram. These instruments are as 
indispensable in the radio laboratory as are T-squares in the drawing 
office and, nowadays, they may be found in many unexpected places. 
Geologists have used them to explore the strata below the surface of 
the earth, trawlers use them to locate shoals of fish, designers of engines 
use them to measure pressures and vibrations, some hospitals use them 
for detecting diseases of the heart and the brain, and they have been 
put to many strange tasks in war. 

Between 1900 and 1930, electrical power and telephone engineers 
were practically the sole users of oscillographs apart from certain 
laboratories and, because cathode ray tubes were not easily obtainable, 
the oscillographs of that period were mainly of an electro-mechanical 
type. In one oscillograph of this type, the image of a pin-point source 
of light is viewed on a translucent screen, the light having passed via 
two mirrors on its way to the screen. These mirrors are so arranged 
that rotation of one mirror makes the spot of light appear to move 
horizontally while vibratory motion of the other makes the spot appear 
to move vertically. In order to make visible the wave-form of an 
electric current, the motion of the mirror which produces vertical 
movement is made proportional to the instantaneous current and the 
horizontal motion is produced by rotating the other mirror at constant 
speed, thus forming a time base. Modern oscillographs employ cathode 
ray tubes and make use of the same idea of motion in two directions at 
right angles, but they have the enormous advantage of being practically 

# As the name implies, the term oscilloscope is only applied when the instrument 
is used for visual inspection of a wave-form or pattern. 



INTRODUCTION 3 

free from the effects of inertia and they are, therefore, able to respond 
faithfully to high frequencies. The improvement in this respect alone 
is of the order of some thousand times. The cathode ray oscillograph 
is also able to perform many tricks which are quite impossible with 
a mirror system. 

In a cathode ray tube, a beam of electrons is focused on a fluorescent 
viewing screen and the resulting luminous spot is moved about either 
by an electrostatic deflecting field or an electromagnetic field or both. 
The spot can respond to very rapid variations of the field because the 
mass of an electron is so very small, and its velocity so very high— the 
electrons take only about one-fifth of a microsecond to pass along the 
length of the cathode ray tube. In the most common arrangement, the 
spot traverses the screen horizontally at constant speed and then flies 
back to the starting-point. This is the time base motion and it is pro- 
duced by a deflecting potential or current which has a saw-tooth 
wave-form. The deflection of the spot at any instant is directly propor- 
tional to the instantaneous value of the deflecting potential or current. 
Because the speed is required to be constant in the forward direction, 
a steadily increasing potential or current is required to produce it, so 
that the sloping edge of the saw-tooth wave-form must be a straight 
line and is termed " linear," i.e. the equation to the curve is * = kt, 
where x is the instantaneous deflection of the cathode ray spot, or the 
instantaneous amplitude of the potential or current which produces it, 
k is a constant and t is the time from the commencement of the trace. 
The equation x = kt is, of course, a linear one. Superimposed on the 
time base motion, there is a vertical deflection in response to the applied 
signal, so that the spot traces out a path which is a graph of the signal 
plotted against time. This graph of the signal is usually referred to as 
the wave-form* of the signal. If the signal recurs at constant time 
intervals the graph is made visible even when it is being traced out at 
fantastic speeds because it is arranged that each horizontal sweep is 
exactly in step with the vertical deflection so that the spot continues 
to trace the same path. At low speeds, the movement of the spot may 
become noticeable although it tends to be concealed by persistence of 
vision and the after-glow of the screen fluorescence. Apart from the 
production of a wave-form diagram or other kind of picture, there are 
many occasions when the time base is employed simply to give a 
measure of time intervals, for counting impulses, or for frequency 
division. In these cases it is not essential to combine the time base with 
a cathode ray tube, but it is often very convenient to do so. 

The basic principle of the majority of these applications of the cathode 

* A wave-form is any potential or current, of changing amplitude which may be 
considered as a function of time, plotted in rectangular co-ordinates. 



4 "TIME BASES 

ray tube is that the luminous spot is caused to move across a screen at a 
velocity which may be either constant or varying in some definite way 
with respect to time, and so to produce a trace on the screen. Signals 
are then applied to the tube so as to disturb the spot and so to indicate 
their arrival. If the normal velocity of the spot is known, then the 
distance between two disturbances on the screen, representing two 
signals, is a measure of the time interval between the signals. The 
disturbance of the spot may be carried out at right angles to its normal 
direction of motion so that it travels along a parallel track while the 
signal is present, or the signal may be applied to control the brightness 
of the spot. Alternatively, the normal motion of the spot may itself 
be interrupted, or otherwise modified, by the presence of the signal 
without disturbing the general direction of the spot motion, so that 
variations in the brightness of the trace are again obtained. In general, 
the time base produces the primary motion of the spot, but it is possible 
to interchange the time base and the signal so that the signal produces 
a regular movement while the time base makes a series of "timing 
marks. " These marks may also be superimposed on an ordinary time 
base trace in order to facilitate the measurement of time intervals or 
to provide a simple means of adjusting the time base trace to a known 
velocity. A modification of this system, in which the time base is 
replaced by another signal, is particularly useful when dealing with 
two related but varying quantities which are thus represented by hori- 
zontal and vertical deflections in such a way that the relation between 
them is expressed in the form of a cyclogram. 

The combination of a cathode ray tube with a time base forms a 
useful tool for the investigation of a very wide variety of problems, but 
the selection of a suitable time base for the particular work in hand is 
most important, and may make all the difference between success and 
failure. The fact that general purpose oscillographs are so satisfactory 
for the observation of recurrent wave-forms has often led to their use 
for work in which a deflection proportional to time is less appropriate 
than one proportional to some other quantity. For example, if it is 
desired to investigate the distortion of an alternating voltage in an 
amplifier valve, we may examine the anode voltage wave-form by 
applying it to the vertical -deflection plates ( Y) of a cathode ray tube 
while the horizontal-deflection plates (X) are connected to a linear 
saw-tooth time base. When the time base recurrence frequency is 
adjusted to synchronism with the frequency of the alternating voltage, 
the wave-form appears stationary on the screen and the degree of 
distortion may be seen. If the time base is now replaced by a distor- 
tionless amplifier whose input terminals are connected across the grid 
and cathode of the amplifier valve under test, the dynamic characteristic 



INTRODUCTION 5 

of the valve will be traced on the screen, and this may be of great 
help in determining the cause of the distortion. Alternatively, if the 
test frequency is a low one, a larger test potential may be used, the 
input to the valve under test being reduced to a suitable value by 
means of a potentiometer and the test potential across the potentio- 
meter being connected to the X deflector plates of the cathode ray 
tube. The possibility of using such a cyclogram should always be 
considered when working on a new problem, but, if it is evident that 
some form of time base is required, consideration should be given to 
the selection or design of a time base circuit which is really appropriate 
to the particular problem. 

Normally, a time base is synchronized by, or arranged to synchronize, 
the signal being examined. For this purpose, it is essential to ensure 
that the time base circuit does not load the circuit under examination 
and distort its output wave-form. If these conditions are not satisfied, 
erroneous conclusions will result. From the point of view of synchroni- 
zation, time bases may be divided into two broad groups, those which 
continue to operate in the absence of a synchronizing signal but which 
are pulled into step by such a signal, and those which are normally 
quiescent and which operate once per signal. The latter type is often 
known as a triggered, single-stroke or externally operated time base. 



CHAPTER II 
TIME BASE WAVE-FORMS 



When a simple saw-tooth time base is used in conjunction with a 
delineating mechanism, such as an oscillograph, a two-dimensional 
image may be obtained which will appear stationary when the wave- 
form being depicted is periodic and when the time base is synchronized 
therewith. The use of a repeating image has the advantage that the 
apparent brightness of the image is greatly enhanced due to the per- 
sistence of vision. Where the wave-form is not periodic only a single 
trace can be used if visual confusion is to be avoided. 

Such arrangements cover most requirements, but occasionally a 
longer trace with equal facility of visual resolution is necessary, and it 
then becomes essential to make use of polar co-ordinates in the form 
of a circular or elliptical time base, or, alternatively, a "time surface" 
or raster may be utilized by the provision of two time bases operating 
at right-angles, one having a high traverse velocity compared with the 
other. Two saw-tooth time bases may be used or one may be circular; 
for instance, the use of a circular time base trace whose diameter is 
controlled by a radial time base results in a spiral or a radial time base 
dependent upon whether the circular component has a higher or lower 
velocity than the radial component. All these methods result in 
expanding greatly the length of trace which it is possible to obtain 
with a given cathode ray tube, and so enable wave-forms to be examined 
in greater detail or for longer periods per time base trace. 

A cathode ray tube can be used as a delineating mechanism for 
examining and measuring electrical phenomena, provided potentials 
of suitable proportions are available for connection to the deflector 
plates of the tube. Alternatively, if there are available currents which 
can be made to pass through deflecting coil systems placed about the 
neck of the tube, the measurements and observations can likewise be 
made. Where it becomes necessary to relate the changes to be examined 
to time, it is obviously important to cause the cathode ray spot to 
traverse the screen, in accordance with a predetermined time law, at 
right angles to the direction of deflection of the wave-form being 
examined. Generally, this time law is a linear one, but, as already 
mentioned, for certain purposes a sinusoidal velocity, or even a very 
complex one, has to be used. 

A repetitive trace along a straight line must obviously be of a dis- 
continuous nature, due to the dimensional limitations of the cathode 
ray tube screen, and the wave-form required to produce it is therefore 

6 




\*^7ime of trace 



TIME BASE WAVE-FORMS 7 

of saw-tooth form, as shown in Fig. i. Note that the saw-tooth wave- 
form is here shown in an ideal manner in so far as the charging period 
is precisely linear, but that, following standard practice, no attempt is 
made to linearize the fly-back portion since nothing is gained by doing 
so. The change of regime from the slow traversing period to the rapid 
fly-back period, and vice versa, must be brought about by a switching 
mechanism. Upon completion of one traverse of the screen, the spot 
is therefore caused to return to the starting-point so rapidly that it 
cannot be seen, or so that, at the worst, its brightness is so far reduced 
with respect to that of the useful traverse that visual confusion is not 
introduced. The return trace can alternatively be made invisible by 
the application of a "blacking-out" signal to the grid or cathode of 
the cathode ray tube by 
utilizing a differentiating 
circuit to perform this 
function. 

It is important to note 
that, provided saturation 
of the cathode ray tube 
screen does not occur, 
the apparent brightness 
of each portion of a re- 
peated image is inversely 

proportional to the screen traverse velocity of the spot at the point con- 
cerned. This, of course, is true only so long as the beam current is 
constant, i.e. so long as the potential difference between the cathode 
and modulator grid of the cathode ray tube does not vary. 

It is sometimes required to provide a time base trace which traverses 
the screen, flies back, and then waits until a given signal before pro- 
ceeding once more to traverse the screen. Such a trace may be required 
where it is necessary to examine, in detail, events which last for a short 
time and which recur at relatively long or irregular time intervals. 

For television purposes, two time bases are used to deflect the beam 
horizontally and vertically. In England, the normal practice is to scan 
the tube screen horizontally 405 times while traversing the screen twice 
in a vertical direction. Each vertical traverse of 202i horizontal lines 
is displaced by half a line so that alternate traverses are interlaced. 
This enables full use to be made of the available number of lines whilst 
, keeping the repetition frequency high enough to eliminate flicker. Fig. 2 
shows the method of building up the English form of television raster. 
This form of raster is due to the Marconi-E.M.I. Company. 

For the examination of radio frequency wave-forms it is usual to 
employ a radio frequency sinusoid of lower frequency as a time base. 



— / /me >■• 

Fig. i. Linear saw-tooth wave-form 



8 



TIME BASES 



This sinusoid is generally obtained from a tuned circuit so that har- 
monics are suppressed. Since the cathode ray tube deflector plate 
capacitance becomes part of the tuned circuit capacitance, it has no 
deleterious effect upon the efficiency of operation. 

When examining a wave-form on a cathode ray oscillograph, using 
a saw-tooth time base, it is often possible to obtain all the information 
that is desired about the signal wave-form without knowing either the 
velocity or the repetition frequency of the trace, but sometimes it is 
necessary, or advantageous, to calibrate the time base sweep so that the 
time axis can be scaled. Calibration may be effected by applying a 
time-marker wave-form in place of the signal. The frequency of the 

time-marker may be ad- 
justed to be an exact 
multiple of the time base 
repetition frequency, or a 
quenched time-marker os- 
cillator may be controlled 
by the time base (see page 
142 et seq.). 

There are various methods 
of calibrating the trace more 
or less continuously, with- 
out disconnecting the signal, 
so that a calibrated time 
scale is made to appear on 
the screen. For example, one may insert an electronic device to inter- 
change the signal and time-marker wave-forms on alternate sweeps. 
The time-marker is then seen on the screen in the form of a scale, 
superimposed upon the signal pattern. 

A form of time-marker which is often more satisfactory is obtained 
by applying the marking wave, which may be obtained from a high- 
speed time base or oscillator giving an output in the form of a pulse of 
short duration, locked to the normal time base, as a deflection of the 
second trace of a double beam cathode ray tube. Such a tube has been 
developed by Fleming- Williams.* The cathode ray beam is split into 
two sections by means of a small plate which is immersed in the beam 
and connected to the final anode. The beam then forms two focused 
points which may be caused to traverse the screen synchronously in one 
direction, but which may be separately controlled in the other direction. 
In another system of continuous calibration, the timing wave -form 

* B. C. Fleming-Williams: "The Double-Beam Cathode Ray Oscillograph," 
Electronics and Television and Short-Wave World, 1930, 12, p. 457. The Double- 
Beam Cathode Ray Tube is made only by A, C. Cossor, Ltd., of London. For a 
description of the tube, see p. 26 1 . 





~^^k 



K 

Fig. 2. Marconi -E.M.I. Television System. 
Diagram showing interlaced scanning 



TIME BASE WAVE-FORMS 9 

is used to control the instantaneous value of the beam current and so 
to produce a series of bright spots at regular time intervals along the 
trace. This method is especially useful in calibrating a cyclogram where 
the absence of any precise indication of the time dimension is sometimes 
a disadvantage. For this purpose one may employ a square-wave time- 
marker in which the mark-space ratio is unity, so that the trace is divided 
into alternate bright and dim sections of equal time duration. 

The time-marker system is elaborated for the production of television 
images in the sense that the time-marker wave-form represents the 
light and shade in the picture and requires to be accurately reproduced. 
It should be noted, moreover, that, even in this case, the time traverse 
is virtually at right angles to the light and shade modulation which is 
here applied as a variation of beam intensity. 

A family-tree of time bases is given on page 10 which shows the 
different methods of application of the time base deflection and the 
resulting forms of time base. It should be noted that, although electro- 
magnetic and mixed deflection methods can be used to produce different 
forms of time base trace, the diagram does not duplicate the information 
which is given under the heading of "Electrostatic Deflection." It 
should also be noted that, under the latter heading, an inductive time 
base is possible but it is, in general, an unsatisfactory device. 
Inductively linearized time bases of the electrostatic* and electro- 
magnetic f types are described later in the book. 

The capacitive form of saw-tooth time base comprises three devices : 

(a) A charging device, which consists of a condenser and a current 
regulating device. 

(b) A discharging, or fly-back, device. 

(c) A switching device. 

In general, the discharging and switching actions are brought about 
by the same device. 

The various types of current-regulating and fly-back devices are 
shown in the diagram, while the fly-back devices are further subdivided 
into self-operated and externally operated devices. 

It will be seen from this diagram that a number of combinations 
are possible, but this book can only deal with a few arrangements. It 
should therefore be understood that, where a particular device has been 
described as forming a part of one time base, it can very often be 
applied with equal success to other time bases. This is particularly so 
with regard to the charging (current regulating) and discharging (fly- 
back) devices used in connection with capacitive saw-tooth time bases. 



* See page 25. 



f See page 124. 



IO 



TIME BASES 



r. 
W 

=3 

F 



2 

W 
P 

o 

h 
u 



— o ™ 5 

•o « p t3 

n C i 5 

J Wuce 

— - - « 



1 

I 

8 ** 



o o 
t2"5> 

11 



o « c 

g g s 

C3 £5 *— 

w o » 8 
> 3 £ 

O s .s 

c «;-c a 

I— I— < IS BJ 



« « 



m c 
-»S- 






• u c 

3.E 
T3 . 
C 



c v 

U 4) » 



•5 a 

%■ 03 



S C u 



a 


1! 


> 




T3 


E 


W 


— a — 




o 


*3 


<A 








u 


S 


CO 


#^ 


to 


*j 




S 



— S.SPK — 



Oh 



— . C 

« i -j 

c 3 h 

C U 3 



rf q c 



- H ■/ 

>Ja .t; 



"u OtJ 



a 3 



QQ 



J3 

3 

h 

I. 

■S 

i 

ra 
U 



03 



J 

-x o «■ 

12 to <n 

Sg-a 



-I 
I 



o^ 






iff J 
Hi a. 






2 £ 

O '£ 

— « 



CO 



§1 






4> « 

£ u 
co£ 



a 

CO 




. 1 
Circular 
trace 




Stiaight- 
line 


l 










> 

■8 

.£ 



1* 



--S 



CO w 



.3 w 

I* 

c <r> 

U o 



w _"2 S_ 



3 



y c 

i *j o 

"2 33 
o 5 S 

CUJ3 to 



CO 



-■? tt — 

T3"5 



v. 



o 



u .- 



CO -a 



2 > o „ 

« M t B 

-g-.S — « S 

If 1 s 

few CO'C 

■■j u 
Qc ♦* ^ o 

,«^ S-Js 



- - ■ - 









CHAPTER III 
TYPES OF TIME BASE 

There are two very distinct types of time base, viz. those operated from 
an A.C. supply and those operated from a D.C. supply, 

TIME BASES OPERATED FROM AN A.C. SUPPLY 
A.C. operated time bases are of two main forms, the sinusoidal and 
the circular time base. Variants of these are the spiral and the radial 
forms of time base which are considered in greater detail in Chapter VI. 

Sinusoidal Time Bases 

The sinusoidal time base is obtained by the simple application of 
a potential of sinusoidal wave-form to one pair of deflector plates of a 
cathode ray tube or by the passage of a sinusoidal current through an 
electromagnetic deflecting system placed about the neck of the tube 
in such a way that the flux passes uniformly across the neck at right 
angles to the direction in which the spot is required to move. 

For many purposes the return trace of a sinusoidal time base is 
disadvantageous and should be blacked out or deflected off the screen. 
For pressure- volume measurements on engines, however, a complete 
sinusoidal trace is sometimes useful. In this application, the measure- 
ment of time is of secondary importance, and a cydogram, relating 
cylinder pressure to piston displacement, is more appropriate. 

Since a small portion in the centre of a sinusoidal traverse is charac- 
terized by an almost constant velocity, it may be used to provide a 
practically linear trace provided the deflecting potential or number of 
ampere turns is sufficiently high. The large amplitude sinusoidal trace 
(equal to many screen diameters) is, however, only of use where a 
relatively small proportion of the total time requires examination because 
only a portion of it remains visible on the screen. 

Circular Time Base 

The method of application of a sinusoidal potential to the deflector 
plates of a cathode ray tube in order to produce a circular time base 
was described by Lilienfeld,* and is shown in Fig. 3, the necessary 

1 
condition of operation being that R = —-. 

coC 
Lilienfeld's time base functions by producing two potentials, which 
are 90 out of phase with one another in time, applied to deflection 
plates which are spatially 90 apart, so that a circular trace results if 

* E. Lilienfeld: German Patent. 



12 



TIME BASES 




Fig. 3, Lilienfeld's circular time base 



the deflectional amplitudes are equal and if harmonics are not present. 
If the deflectional amplitudes are unequal, or if R is not equal to 

— — , the trace becomes elliptical. 

If a circular time base trace is to be employed for the measurement 
of time intervals it becomes important to ensure that the trace is 

actually circular, since only 
then will the velocity of the 
spot be constant. As the sen- 
sitivity of the pair of plates 
nearer the gun is greater than 
that of the other pair, it is 
advisable to connect the for- 
mer pair across a part of the 
resistance R, which now 
becomes (at least in part) a 
potentiometer. By adjusting 
the potentiometer, it is possible to obtain equal deflections from the 
two pairs of plates. Moreover, since distortion will occur, due to the 
use of unbalanced deflection with a normal tube, the use of an anti- 
trapezium tube is necessary unless push-pull deflection methods are 
adopted. The correct setting-up procedure is, of course, to adjust the 

variable resistance until R = -_ and then to adjust the potentiometer 

until circularity of the trace is obtained. 

Another method of obtaining a 
circular time base trace is to pass a 
sinusoidal current through a pair of 
coils symmetrically disposed about the 
neck of the tube. The pair of plates 
whose line of deflection is parallel with 
the field produced by the coils is then 
connected across the ends of the coils. 
In general, an elliptical trace will re- 
sult, since the current through the 
coils is nearly 90 out of phase with the 
potential across the coils. Circu- 
larity of the trace can be obtained by 
adjustment of the distance between 
the coils. The method, which is 
shown in Fig. 4, is also applicable to tuned circuits, but the value 

— should be high in either case, so that the current in the coil lags as 
R 



^ 



\ h 



T 



Fig 



4. Circular time base 
(semi-magnetic) 



TYPES OF TIME BASE 



*3 



nearly as possible by 90 behind the applied potential. A true circle 
will not be obtained unless this 90 is realized, but a close approach 
is possible. 

It is sometimes required to know the direction of rotation of the 
trace; this information may be obtained by a consideration of the circuit 
arrangements as shown in Fig. 3. Assume that the deflector plates 
appear in the drawing in the positions which they occupy when viewed 
through the screen of the cathode ray tube. Then, since the potential 
across the resistance R leads by 90° with respect to that across the 
condenser, the deflector plates will receive a positive potential in the 
following sequence: Y 2 X x Y x X 2i and the trace will rotate in a 
clockwise direction. 

If the condenser C is replaced by an inductance having a negligible 
resistance, the potential across R will lag by 90° with respect to that 
across the inductance, and the trace will rotate in an anti- clockwise 
direction. 

There is no need to describe these deflecting methods further at the 
moment, other than to point out that, for electrostatic deflection, the 
movement of the spot is in phase with the deflecting potential, that is 
to say, when the potential between a pair of plates is zero the spot lies 
on the centre line of the deflectional axis of the pair. In the electro- 
magnetic case, the deflection is in phase with the current in the deflecting 
coils; thus, for a deflecting system in which the circuit resistance is 
negligible, the spot lies on the centre line of the tube when the potential 
is a maximum. The movement of the spot then lags by 90 behind 
that of the supply potential, since the position of the spot is determined 
by the instantaneous amplitude of the current in the coil, and this also 
lags by 90 with respect to the potential across the coil. 



TIME BASES OPERATED FROM A D.C. SUPPLY 

Most time bases which are operated from a D.C. supply are intended 
to provide a linear saw-tooth wave-form, i.e. for the period of one 
trace they are linear in form and constant in velocity as distinct from 
the sinusoidal time base which is linear in form, but has a varying 
velocity and the circular time base, which is not linear in form, but is 
constant in velocity. The ideal to be aimed at in obtaining a linear 
saw-tooth time base is to provide a potential, or a current, which rises 
or falls linearly with respect to time, but this is not easily obtained by 
direct means, and resort must usually be had to various tricks in order 
to linearize or to compensate exponential characteristics. 

Linear saw-tooth time bases may be subdivided into those in which 
the action is obtained by: 



H 



TIME BASES 



(a) the change of potential across a condenser upon the application 
of (ideally) constant direct current via a resistance (Capacitive 
Time Bases), 

(b) the change of current through an inductance upon the application 
of a constant direct potential (Inductive Time Bases), 

All time bases of types (a) and (b) require a switching mechanism 
to change the regime from the traversing to the fly-back condition. 
Furthermore, time bases of these types usually employ separate charging 
and discharging circuits and, since the impedance of the discharge 
circuit is necessarily very low by comparison with that of the charging 
circuit, due to the small time which is allowed for the discharging 
process, it is, in practice, only necessary to switch the discharging 
circuit. The charging circuit is thus permitted to function continuously. 
The discharging arrangement generally consists of one or more valves 
which are switched by the time base itself or by an external circuit. 
These discharging arrangements may employ hard or soft valves, i.e. 
high- vacuum or gas-filled valves may be used. 

Self-operated circuit switching is usually brought about by the use of : 

(A) Biased thyratron or gas- discharge valves. 

(B) Multivibrators and other trigger circuits. 

(C) Blocking oscillators. 

Switching arrangements of type (A) are implicitly of the soft valve 
type, while those of types (B) and (C) are of the hard valve type. 
A further important classification remains, viz : 
(i) Self-operating time bases. 
(2) Externally controlled or triggered (single-stroke) time bases, 

Capacitive Time Bases 

Capacitive time bases depend on the fact that if a steady direct 
potential is applied across a condenser and resistance in series, current 

flows into the condenser which thus acquires a potential across it, 

where Q is the charge upon the condenser in Coulombs and C is the 
capacitance in Farads. Unfortunately, from the point of view of 
producing a linear saw-tooth time base, the rise of potential v across 
the condenser is exponential in shape, the potential at any instant being 

El i-e BS V where E is the supply potential, C the capacitance, R the 

resistance, and t the time in seconds after the commencement of 
operation (see Appendix II). It is sometimes required that a saw-tooth 
time base shall have an exponential wave-form, and, in this case, no 
disadvantage accrues. 









TYPES OF TIME BASE 



15 



The capacitive type of time base has passed through many stages 
of development by a large number of workers and is probably the 
most important form of time base in use to-day. This is particularly 
true where the possibility of obtaining a large range of operating 
frequency is involved as in the case of those time bases which are 
intended to be used for a number of different purposes such as are 
required for general investigation and measurement work. For this 
reason, this type of time base will be discussed in fairly complete detail. 

Inductive Time Bases 

The general principle upon which the inductive time base depends 
is that, if a direct potential be suddenly applied to a pure inductance, 
the current rises proportionally with respect to time and therefore this 
current can be utilized to provide a deflection of the spot in a cathode 
ray tube. If a resistance of negligible proportions is included in series 
with the inductance, the potential across it may be amplified and applied 
to the deflector plates of a cathode ray tube. Preferably, the inductance 
should itself be used for the purpose of providing electromagnetic 
deflection. These are discussed on page 95 et seq. 

SOFT VALVE TYPES OF LINEAR TIME BASE 

Capacitive Time Bases 

The earliest attempt at the construction of a linear saw-tooth time 
base was made by Pearson and Anson* in the year 1922. The author 
believes that this instrument appeared as a result of the development 
of the Anson relay in which a neon tube was used for signal-shaping 
purposes in conjunction _j_ q_ 
with telegraph receivers. 
The neon tube is a two- 
electrode valve, filled with 
neon gas at a low pressure, 

in which neither of the "J"'"' R 

electrodes is heated. When — °- 
a potential, which is de- 
pendent upon the gas 

., • 4_ £ Fig. 5. Simple Neon time base 

pressure, the proximity or 3 

the electrodes, the material of which the electrodes are made and their 

surface condition, is applied between the electrodes the gas becomes 

ionized. This potential, known as the striking potential, is normally 

about 130 volts. When the potential falls to about 100 volts the gas 

deionizes. 

* H. St. G. Anson: British Patent 214754. S. O. Pearson and H. St. G. Anson: 
"The Neon Tube as a Means of Producing Intermittent Currents," Proc. Phys. Soc, 
of London, 1922, 34, Part V, p. 204. 



D.C 

Supply 




I 



Cz 



-oOut 



-oEarth 



C 3 



-oSynch. 



i6 



TIME BASES 






The neon time base shown in Fig. 5 consists of a condenser charged 
through a high resistance and discharged by the neon tube when the 
charge upon the condenser reaches the striking potential of the tube. 
The amplitude of the time base output potential is thus limited to the 
difference between the striking and extinction potentials of the neon 
tube — usually a range of about 30 volts, and the charging curve is 
therefore a small portion of an exponential curve as shown in Fig. 6. 
By using a high charging potential and a high value of charging resis- 
tance R x , the degree of curvature can be materially reduced at the 
expense of the potential efficiency of the time base, where the efficiency 
is expressed as the ratio 

potential excursion of the condenser plates 
steady supply potential 

Typical values for the components shown in Fig. 5 are: 

7^=2-0 megohms, 
.R 2 = i,ooo ohms. 

In the time base of Fig. 5 the repetition frequency is, of course, 
controlled by varying either the capacitance of the condenser C % or 

the value of the charging resis- 
tance J? a . If a portion of the 
potential to be examined is im- 
pressed across a resistance R. z 
connected in series with the neon 
lamp, the time base may be 
synchronized with the potential 
under examination. The ques- 
tion of the synchronization of 
a time base is discussed in 
Chapter XL 

In 1924, Appleton, Herd and 
Watson Watt* replaced the charg- 
ing resistance by a saturated valve, i.e. one in which the electron 
emission is limited by the temperature of the filament. As a result of 
this saturation the charging current becomes more constant, and thus a 
closer approximation to a linear rate of charge of the condenser is 
obtained. It is perhaps advisable to use a valve having a plain metal 
filament as a charging valve, although some time bases of this type have 
employed valves having thoriated tungsten filaments. The use of a 
saturated diode renders the time base somewhat sensitive to supply 
voltage variations. 

* E. V. Appleton, J. F. Herd and R. A. Watson Watt: British Patent 235254. 




7TI 

V\l 



V c ■• yo/tac/e change ccrcss Ci 



- 



. trie *— 

FlG. 6. Potential output obtained from 
the time base of Fig. 5 



TYPES OF TIME BASE 



17 



About the middle of 1933 Bedford* and von Ardenne, working 
independently, replaced the diode by a pentode and obtained better 
results so far as the linearity of the charge is concerned. To use a 
saturated diode, or a pentode operating above the knee of the anode 
volts-anode current characteristic, instead of a resistance element in 
series with the condenser, to form a time base charging circuit is to 
replace the (fixed) resistance by one whose value is dependent upon 
the potential across it in such a way that the current flowing through 
it tends to remain constant. The slope of the wave-form of the potential 
across the condenser is proportional, at every instant, to the value of 
the charging current. If, therefore, the variations in the current are 
reduced, the slope becomes more constant, and the curvature of the 
charging characteristic is also reduced. This does not necessarily mean 
that the law has ceased to be an exponential one and, indeed, so long 
as the "constant-current" device is operating on the linear portion of 
the valve characteristic, the law will be exponential. The term "linear 
portion of the valve characteristic" defines that portion where the 
relation between current and voltage is a straight line over the working 
range. This is approximately the case with a pentode when the anode 
potential is above approximately 50 volts. The potential efficiency of the 
pentode charging circuit is high because the charging process may be 
continued until the anode potential has fallen to about 50 volts. The 
potential across the condenser will then be 50 volts less than the rail 
potential, which for a 500-volt supply is a charge of 90 per cent of the 
available volts. With the simple resistance charging circuit the potential 
efficiency depends entirely on the degree of linearity to be reached. 
Thus, if the rate of charge is to be allowed to fall by as much as 
10 per cent, the maximum potential across the condenser will be 
only io per cent of the rail potential, or 50 volts for a 500-volt supply. 
The effect of using a pentode in this way is to provide approximately 
the same result as would be obtained by using a resistance having a 
value equal to the slope impedance of the valve and a supply potential 
equal to RJ a where R a is the slope impedance f and I a the anode current 
in amperes. It will thus be obvious that the value of the impedance of 
the charging circuit is high. 

It is important to note how the relationship between charging time 
and discharging time is affected by the circuit constants. The charging 

* L. H. Bedford and O. S. Puckle: "A Velocity-Modulation Television System," 
Journ. I.E.E., 1934, 75, p. 63. . « . . . 

t The slope impedance (A.C. resistance) of a pentode is found by noting the 
increase of anode current resulting from a specified increase of anode potential, the 
screen and grid potentials remaining constant. 

,„ , change of anode potential 
Slope impedance (i?«) = ^^ of — ode cu ^ nt ' 



i8 



TIME BASES 



time is determined by the current flowing into the condenser and the 
potential to which the condenser is to be charged. These are likewise 
determined by the supply potential, the charging impedance, the capaci- 
tance of the condenser and the striking potential of the discharge tube. 
The discharging time is determined by the discharge current (which is 
dependent upon the impedance of the discharge circuit) and the capaci- 




tance of the condenser. Thus, the ratio 



charging time 



is dependent 



discharging time 

. . average value of I d , , . . . . . . , , , 

upon the ratio 5 ■ where I d is the discharging and J c the 

average value or i c 

charging current. The average values here referred to are those for 

the period of one charge and for the period of one discharge. These 

ratios are increased by decreasing the charging current and by making 

the impedance of the discharging circuit as small as possible. There is, 

however, a limit to the amount by which the condenser charging current 

can be reduced since it must be large compared with any leakage 

currents through, or across, either the condenser or the charging valve, 

including the current through any leak resistances connected across the 

deflector plates of the cathode ray tube. 

If the anode current is not large by comparison with the leakage 
currents in an otherwise constant-current charging circuit, the charging 
current will contain an appreciable component which is dependent 
upon the instantaneous potential across the condenser; such a condition 
will, of course, result in non-linearity of the trace, even if the leakage 
path is in series with a further condenser. This condition exists only 
because the impedance of the charging circuit is high and leakage is 
therefore important. If the charging circuit is of low impedance, parallel 
leakage paths generally have little effect. Some circuits which have a 
low value of output impedance, as a result of the use of negative voltage 
feed-back, are described in Chapters VIII and IX. 

From the foregoing remarks it will be evident that the linearity 
obtainable with a pentode depends primarily on the equivalent potential 
which is the product of the slope impedance and the anode current, 
i.e. R a I a . If the pentode is operated with a low screen potential, or a 
negative grid bias, the slope impedance will be raised and the anode 
current will be reduced. The product of the impedance and the anode 
current remains, however, approximately the same, so that it appears 
at first sight that one could use a very small charging current and a 
very small condenser in order to obtain a relatively rapid fly-back. The 
effect of the leakage, however, is to set a limit to the application of this 
idea and, in practice, it is found difficult to retain a high degree of 
linearity unless a substantial charging current (say, at least, i milli- 
ampere) is employed. 



TYPES OF TIME BASE 



19 






The time base, in the state which it had reached in 1928, was still 
somewhat unsatisfactory in that the potential output, and hence the 
amount of deflection obtainable on the cathode ray tube, was not very 
large. Since a contemporary tendency was towards the use of high 
anode potentials for the gas-filled cathode ray tube, the failure to 
produce sufficient output potential began to become of importance. 
At about the same time Hull* introduced the thyratron, which im- 
mediately improved the potential efficiency, due to an increase in the 
separation between the striking and extinction potentials, and so pro- 
vided a considerable impetus to the use of the cathode ray oscillograph. 
The combination of the new valve with the constant-current pentode 
a few years later seemed, at the time, to be a final solution. The thyra- 
tron not only has the desired large separation between the striking and 
extinction potentials but this separation is very easily controlled at will. 

The thyratron is a valve containing a hot cathode and a control grid, 
together with a small quantity of mercury inserted after evacuation of 
the bulb. The heat from the cathode vaporizes some of the mercury 
and, since the vapour pressure is dependent upon the bulb temperature, 
the ambient temperature of the room in which it operates has an 
appreciable effect upon the density of the vapour and, hence, upon the 
behaviour of the tube. A type of threshold effect comes into play when 
the anode and grid potentials are sufficient to cause a certain value of 
anode current to pass. This ionizes the vapour in the anode-cathode 
space, a cloud of positive ions surrounds the grid, so causing it to lose 
control, and an arc forms between the anode and cathode with a 
potential drop of from 15 to 18 volts. It will be appreciated that if 
the grid bias and anode potentials are such that insufficient anode 
current flows to initiate this violent ionization, the valve behaves rather 
like a hard valve biased almost to cut-off. Thus, as the condenser 
charges, the anode potential rises until the potential gradient is sufficient 
to ionize the vapour so that a heavy current passes and rapidly discharges 
the condenser, whereupon the cycle recommences. 

When the vapour is in the ionized condition, it can be deionized 
only by reduction of anode potential and not normally by reduction of 
grid potential. In this way, adjustments of the control grid potential 
may be used to control the initiation of ionization, but not to control 
the cessation. 

A normal method of use is to connect the condenser between the 
anode and cathode of the thyratron, with a small resistance in series, 
and to adjust the grid bias potential so as to obtain the desired striking 
potential. The ratio of the striking potential to the grid bias potential is 

Fl* ?"■ ^' Hull: "Hot-Cathode Thyratrons." Part I. "Characteristics," General 
atectnc Review, 1929, 32, p. 213. Part II. "Applications," ibid., 1929, 32, p. 390. 



20 



TIME BASES 



known as the control ratio, the value of which generally lies between 
about 20 and 40, Adjustment of the grid bias potential enables the 
striking potential to be set to a predetermined value which may be as 
high as 1,000 volts. The potential to which the condenser charges 
and, therefore, the amplitude of the motion of the spot on the cathode 
ray screen is thus placed under the control of the operator and the 
maximum value of this potential is many times greater than can be 
obtained when using a neon tube as the discharge device. Furthermore, 
in the case of a neon tube, synchronization can only be obtained by 
adding the synchronizing potential in series with the potential across 
the condenser. In the case of a thyratron, however, the controlling 
effect of the grid may be utilized to synchronize the action of the time 
base with the wave- form under examination. If this wave-form, 
adjusted to have a suitable amplitude, is applied to the grid it will 
initiate the discharge at the correct moment. In this way, the valve 
amplification ensures that synchronization can be achieved with a much 
smaller loading effect upon the circuit under examination than is the 
case with a neon tube time base. 

The improved synchronization and the great increase in deflection 
amplitude obtained with the thyratron, plus the improved linearity due 
to the use of the pentode, provides an instrument which meets many 
requirements. One requirement which, however, is not met is that of 
a time base having a very wide frequency range, since the thyratron 
time base is normally limited to a maximum repetition frequency of 
approximately 40 or 50 kilocycles per second. This frequency limita- 
tion of the thyratron time base is due to the fact that as the capacitance is 
reduced, the charging current must also be reduced. If the current is 
not reduced, the discharge fails to extinguish, for very small capaci- 
tances, when the condenser has been discharged (see Appendix III). 

In an attempt to improve upon the thyratron as a discharging device, 
Bedford replaced mercury vapour by neon and other inert gases, and 
this type of valve became known as a gas-filled triode or gas-filled relay. 
Some, but relatively little, improvement was obtained in the maximum 
repetition frequency, but great improvement was obtained in stability. 
The reason for the improved stability is that, in the case of a thyratron, 
there is a small quantity of liquid mercury in the bulb, and changes of 
temperature alter the amount of vapour. In gas-filled valves, the 
quantity of gas remains constant except for a secondary effect due to 
the gradual adsorption of the gas into the metal parts of the valve, but 
this process takes a considerable time. 

The circuit of a typical thyratron, or gas-filled relay, time base is 
shown in Fig. 7. By using a switch to select one of a number of con- 
densers and by varying the value of the charging resistance or, as in 




TYPES OF TIME BASE 



21 



this case, the screen potential of the charging valve, it becomes possible 
to obtain a continuous variation in the rate of operation of the time 
base over quite a reasonable frequency scale. 
The following typical values may be employed for Fig. 7 : 



H.T.=35o volts Q|=I t*F. 

V ± Screen potential 20-150 volts i? 2 =500 ohms. 

It is important to note that, with regard to Fig. 7, the time constant 
C 3 #3 should be large compared with the time duration of the slowest 
time base trace which the instrument is required to produce. 

-. — 1 ♦— —oHX+ 

£z£z 

oSynch. 



The 



Velocity, 




oOut 



0//.Z- 



Fig. 7. Thyratron or gas-filled relay time base 



process which occurs is as follows: each time the condenser C 2 is 
discharged, the anode current of the valve V 2 passes into the condenser 
C 3 and gives it an increment of charge. While the condenser C 2 is 
again being charged, and so producing the next time base trace, the 
condenser C 3 discharges slightly through R 3 . The effective bias on 
the grid of F 2 is that of the potential across C 3 when the gas in the 
valve again becomes ionized, hence the necessity for making the time 
constant C 3 J? 3 large in order that the potential across the condenser 
shall not fall appreciably during the charging period. The bias on 
the grid of V % is equal to the value of R B times the average value of the 
current through it. This current is equal to the average value of 
the anode current of V x and is easily measured by connecting a 
nulliammeter in the anode circuit of that valve. The striking potential 
therefore depends on the adjustment of R x and of R z so that it is not 
possible to control the repetition frequency of the time base without 
introducing a change in the amplitude of the trace. When the rate of 
charge is increased by an increase in the screen potential of V lf the 
condenser C 2 must, of necessity, be discharged more frequently in a 
given time and, hence, the mean anode current in V 2 increases. The 



22 



TIME BASES 



TYPES OF TIME BASE 



23 



value of R s must therefore be reduced if the bias potential is to remain 
the same as before. 

It should be noted that when the time base is first switched on, the 
leakage current through the condenser C 3 , if it is an electrolytic one, 
is likely to be large and this leakage is, in effect, a low resistance con- 
nected in parallel with R s . As a result, the effective value of 2? 3 is 
reduced and the bias potential will be less, so that C 2 will charge to a 
smaller potential. Thus, the time base trace length will be shortened 
and the repetition frequency will be high until the condenser plates 
form to the working potential, after which the leakage current will fall 
to a normal value. The resistance of the circuit will take a few minutes 
to settle down to its normal value. 

It should be borne in mind that thyratrons and gas-filled valves are 
very easily damaged by excessive currents. It is, therefore, essential 
to limit the current in both the anode and grid circuits. The anode 
current must be limited by a resistance R 2} which is, at least, equal to 

E 

-v, where E is the maximum voltage to which the condenser is likely to 

be charged and / is the instantaneous peak current rating of the parti- 
cular valve in use. Typical time base thyratrons and gas-filled valves 
will safely carry 200 to 500 milliamperes in the anode circuit. This 
resistance should not be made higher in value than is necessary since, 
otherwise, the discharge may be difficult to extinguish, particularly 
with the smaller values of capacitance in the main charging circuit (see 
Appendix III, page 313 et seq.). 

An alternative method of limiting the anode current is to connect 
a separate inductance in series with each condenser in the anode circuit 
of the discharge valve only, the resistance being omitted. The required 

CE 2 
value of the inductance is given in each case by the equation L =— ™ » 

where L is in microhenries, C in microfarads, E in volts, and / in 
amperes. 

If no grid resistance is used, the grid current is likely to rise to an 
excessive value during the ionization period and, if this current flows 
for a long time such as occurs when discharging a condenser having 
a capacitance of 1 microfarad or more, the grid may become heated 
to such an extent as to cause it to emit electrons. If this happens, the 
condenser will not be charged to its correct potential because the 
electrons emitted from the grid will ionize the gas at a lower anode 
potential. Thus, if the maximum bias potential is likely to be, say, 
15 volts, the value of R 4 should be of the order of 5,000 ohms so that 
the grid current will be limited to 3 milliamperes. It is not advisable 
to make the grid circuit resistance higher than is necessary, although 



the actual value is not critical. The operation of the time base is, 
however, likely to become erratic if the grid circuit impedance is very 
high. In this case, the grid potential may be altered by pick-up from 
the mains or from other parts of the circuit by stray capacitance, in 
which event successive traces will be of different lengths due to the 
continually varying value of the grid potential. In some instances, it 
may prove necessary to increase the value of R 4 in order to avoid 
overloading the source of the synchronizing signals. Preferably, a 
resistance may be placed between C 4 and the "synch" terminal or, 
alternatively, the value of C 4 may be reduced to bring about the same 
result. Each of these changes also has the effect of avoiding a large 
value of grid current in the event of the impedance of the synchronizing 
source being very low. The extra resistance in series with C 4 may 
require to be shunted with a small condenser of a few picofarads in 
order to preserve the steepness of the synchronizing wave-form. 

Another reason for adding a resistance between the condenser C 4 and 
the synch terminal, or for making the value of C 4 small, is that, if this is 
not done and if the impedance of the synchronising source is low, the 
alternating potential of the grid will be held near the potential of the 
negative supply rail and therefore the bias control resistance R 3 will 
not be able to control the amplitude of the time base trace. 

An alternative form of the bias circuit shown in Fig. 7 is obtained 
by short-circuiting R 3 and C 3 and by connecting the lower end of /? 4 
to a potentiometer across the high-tension supply rails. The striking 
potential then becomes independent of the adjustment of R v When the 
discharge valve strikes and discharges the time base condenser, the 
grid and cathode of the valve are held at approximately similar 
potentials so long as the valve is in the conducting state. The potential 
drop across /? 4 may, in this case, momentarily reach a value equal to 
nearly the full rail potential, and the value of /? 4 should be chosen so as 
to limit the grid current to a few milliamperes in this condition. A 
condenser of adequate size should be connected between the potentio- 
meter slider and the negative rail to prevent the appearance of ripple 
on the grid. 

A further point to be borne in mind when constructing a thyratron 
°r gas-filled relay time base, is that the potential between the heater 
and cathode must not exceed a few volts. The use of a separate heater 
winding on the mains transformer is necessary, and this should usually 
he screened from the other windings to prevent hum. One of the heater 
terminals may then be connected to the cathode of the thyratron. 

All power supply circuits should have a low output impedance at 
he lowest frequency of the circuit which they are designed to drive. 



Thi 



s means that the smoothing circuit must, in general, end with a 



2 4 



TIME BASES 



parallel condenser which must be of sufficient capacitance to ensure the 
low impedance required. As a result, the positive and negative supply 
rails are at substantially the same potential from an A.C. point of view, 
and the condenser C 2 , in Fig. 7, may therefore be connected between 
the anode of V x and the negative supply rail if desired. The circuit 
then behaves, to all outward appearances, exactly as before and, in the 
A.C, sense, it is precisely the same as if the condenser were still 
connected to the positive rail. 

This time base, and most other capacitive time bases, may, of 
course, be reversed as shown in Fig. 8, in which the condenser C 2 
and the discharge valve F 2 are at the negative end of the circuit and 



OKT.+ 



oOut 





—=—o$ynch 



*>MT. m 



Fig. 8. Inverted thyratron time base 



the charging resistance is at the positive end. Such an arrangement 

cannot easily be made to function satisfactorily when using a pentode 

charging valve because it is difficult to fix the screen potential with 

respect to that of the cathode since the potential of the latter is rapidly 

varying. 

The average rate of rise of potential across the condenser in the 

anode circuit of a charging valve is calculated from the equation 

I x 10^ 
U e = — , where U e is in volts per second, I e the average anode 

current in amperes, and C the capacitance in microfarads. If the 
condenser is connected directly to a cathode ray oscillograph, the length 
of the trace and, therefore, the potential variation V across the condenser 
may be measured (see page 269). Having obtained the value of the 
potential variation across the condenser and the rate of rise of potential, 
it becomes possible to find the duration of the trace. This is, of 

course, equal to - , in seconds. 

The duration of the fly-back can be determined by applying the time 
base to the Y deflector plates of a cathode ray tube and a separate 



TYPES OF TIME BASE 



25 



linear saw-tooth time base to the X plates. If the second time base is 
adjusted to show two saw-tooth waves of the time base under examina- 
tion, the time duration of the fly-back may be measured by comparison 
with that of the charging period which is already known. 

The repetition frequency of a time base is best measured by com- 
parison with a beat-frequency oscillator. It should be remembered 
that, if the time base is synchronized, the repetition frequency will 
be altered. 

Lewer and Dunham* have described many other uses of the thyratron. 

Blumlein's Thyratron Time Base 

In 1932 Blumleinj developed a form of time base which operates 
with a thyratron although the discharge circuit is modified by the use 
of a series inductance. The circuit of this time base is shown in Fig. 9, 



1 — SW$&~ 
12 



O 



c 2 



I 



-oOt/t 






SffitiP WVSr 



-o/f.T+ 



r 
Mia 



nc* 



OHI- 



Fic. 9. Blumlein's thyratron time base 

from which it will be seen that it consists of a condenser which is 
charged by a substantially constant current and discharged through a 
thyratron. A novel feature of this circuit is the combination of: 

(a) a large inductance employed in the charging circuit for improving 
the linearity of the charge, and 

(b) a small inductance used in the discharge circuit. 

The condenser C x charges through an inductance L ly having a value 
of the order of 30 henries, in series with a high resistance, so that 
the current through the condenser is substantially constant. If the 
thyratron is removed and a steady potential is suddenly applied to the 
circuit, the current executes a damped oscillation at a frequency equal to 

/ I ~~P 



., where R is the value of the charging resistance in series 



2tc 



with L l7 before decaying to zero. This is shown in Fig. 10. During this 

time, the potential across the condenser swings above and below the level 

of the supply potential before finally settling down at this value, as shown 

S. K. Lewer and C. R. Dunham: Gas-filled Relays, published by the General 
Electric Co., Ltd., London. 

f A. D. Blumlein : British Patent 400976. 



26 



TIME BASES 



TYPES OF TIME BASE 



in Fig. ii. Normally, the discharge through the valve interrupts this 
process during the first half-cycle, thus, it is evident that the voltage 
rise is neither exponential nor truly linear, but is actually a small part 
of a sine curve. When the discharge valve is operating the situation is 
complicated by the existence of an initial potential across the condenser 



-Discharge commences here 



27 




Time 
Fig. 10. Charging current wave-form for the circuit of Fig. 9 



-Discharge commences here 
. Appfied potential 




Time 



Fig. 11. Potential wave-form across the condenser C t in Fig, 9 in 
the absence of the thyratron 

C x and an initial current in the inductance L x at the commencement 

of each charging period. The maximum potential across C l exceeds 

that of the H.T. supply potential and may be nearly twice as great, 

A similar action takes place in the discharge circuit. If the discharge 

valve is replaced by a very low resistance, the current rises and falls 

1 
at a frequency approximating to — ,_ _ > since the resistance is 



2*VL 2 Cl 



assumed to be too small to have any appreciable effect upon the 
frequency. This frequency is much higher than that of Fig. 10, 
since the value of L 2 is much less than that of L v Since the valve 
cannot conduct in the reverse direction, the circuit is effectively 
broken at the instant when the oscillatory current is on the point of 
making its first reversal, i.e. at the completion of the first half-cycle. 
At the same time, the potential across the condenser also follows a sine 
curve, but it is out of phase with the current and, at the instant of zero 
current, the condenser is left with a charge in the reverse direction.* 
Fig. 12, drawn on an enlarged time scale, shows the complete cycle 
of condenser potential variations. 

The energy originally stored in the condenser is thus transferred first 
to the inductance and then back again to the condenser, although not 




Fig. 12. Output wave-form for the circuit of Fig. 9 

without some loss due mainly to the resistance of the coil. After the 
cessation of the discharge current, the anode of the valve becomes 
negative, the condenser potential approaching its original value but 
in the reverse direction. The charging current, which is naturally very 
small in comparison with the discharge current, then causes the potential 
on the upper plate of the condenser to become steadily more positive, 
thus producing the time base trace. This brings about the enviable 
result that the electrostatic deflecting potential which is available at 
the condenser terminals, is approximately doubled by comparison with 
that obtained from an ordinary resistance-limited thyratron time base. 
I'or this reason, time bases of this type, and those shown in Figs. 66, 
68 and 70 are known as high-efficiency time bases. 

Since the form of the discharge current is practically that of a half- 
sinusoid, the condenser will be discharged more rapidly for a given 
maximum anode current in the discharging valve than is the case when 
discharging a condenser which has a resistance connected between it 
and the anode of the thyratron. In the latter case, the form factor 

■ , " e reversal of charge when a charged condenser is short-circuited through an 

uetance is easily demonstrated with the aid of a large condenser (say 4 microfarads), 

re' ■? lomentar ^y short-circuiting it through a few feet of wire. In measuring the 

sidual potential with a D.C. voltmeter it will often be found that a reversal of polarity 

M" taken place. 



28 



TIME BASES 



TYPES OF TIME BASE 



29 



is much higher and the maximum current flows for a shorter period 
of the total time. 

A feature of the circuit is the provision of a tuned acceptor circuit 
L 3 , C 2 , which may be connected across the charging condenser to 
eliminate disturbances at half the operating frequency. If this refine- 
ment is omitted a slight displacement of alternate charging strokes is 
apt to appear. Note that, although this time base employs an inductance 
it is not an inductive time base since the element which produces the 
time base trace is the condenser Cj. 

Faudell and White's Thyratron Time Base 

A somewhat similar circuit which provides a push-pull output has 
been developed by Faudell and White.* In this circuit, the cathode of 
the valve has a tuned circuit between it and the negative supply rail. 
The two outputs are then taken from the following points : 

(a) across the condenser C x as in Fig. 9, and 

(b) across the tuned circuit in the cathode of the valve. 

HARD VALVE TYPES OF LINEAR TIME BASE 

Many discharging circuits for time bases have been developed along 
the lines of the relaxation oscillator which consists, in its simplest form, of 
a two stage resistance-capacitance back-coupled amplifier in which each 
stage generally has the same circuit constants and in which the degree 
of feed-back is large. Such circuits are known as Trigger or Flip-Flop 
circuits, and are of very great importance. The earliest and best known 
of these circuits is the Multivibrator. 

Self-Operated Time Bases of the Multivibrator Type 
Abraham and B loch's Multivibrator 

The multivibrator, described by Abraham and Bloch f in 1918, is 
a very important circuit, and is in use for many purposes. In its 
original form, however, it is more commonly used for the generation 
of pulses than as a time base or discharger circuit. 

The multivibrator is often used as a generator of harmonics of a 
synchronizing source for the production of standard frequencies. J In 
this case it is usual to employ either an electrically maintained tuning 
fork or a piezo-electric crystal as the source of the synchronizing signal 

* C. L. Faudell and E. L. C. White : British Patent 479275- 

t H. Abraham and E. Bloch: "Notice sur les Lampes-valves a 3 Electrodes et 
feurs Applications" ("Information on three-electrode valves and their applications"), 
April 19 1 8. Publication No. 27 of the French Ministere de la Guerre. G. W. O. Howe : 
"The Equivalent Circuit of the Multivibrator," Wireless Engineer, 1945, 22, p. 572. 

J H. Abraham and E. Bloch : " Sur la Mesure en Valeur Absolue des Periods des 
Oscillations £lectriques de Haute Frequence" ("On the absolute measurement of the 
period of electric high-frequency oscillations"), Comptes Rendus, 19 19, 168, p. 1105. 



and harmonics up to, at least, the 150th may be obtained by selection 
with the aid of a tuned circuit. Note that the use of an external syn- 
chronizing signal does not necessarily imply external operation. 

The multivibrator, which is shown in Fig. 13, consists of two valves, 
each having the anode coupled to the grid of the other via a condenser, 
and with a leak resistance to earth. The method of operation of the 
circuit is as follows: when the anode current in the valve V x increases 
for any reason, a negative signal is passed to the grid of V 2 , due to the 
increased potential drop across the anode load R s . The anode current 
of V 2 is thus reduced, which results in the grid of V 1 becoming more 

•OM.T.+ 




T 



£3 1C5 



_j — 






.1. 

■¥■ 



.1. 



J L 



O/f.I- 



FlG. 13. Abraham and Bloch's multivibrator 

positive, thus increasing the anode current of this valve still further. 
When the overall gain of the circuit exceeds unity, the effect is a cumu- 
lative one and the anode current in V 1 rapidly reaches a maximum 
while the anode current in V 2 is cut off. This circuit condition then 
remains until the charge in the condenser C 1 has leaked away through 
V x and R % sufficiently to permit anode current to flow once more in the 
second valve. When this occurs, the cumulative effect again takes place, 
but in the reverse direction. 

The multivibrator, therefore, has two unstable limiting conditions 
which occur when either of the valves is at cut-off and the other at 
zero grid potential. 

The following typical values may be employed for Fig. 13 : 

Ri=R% 1 -o megohm H.T. 250 volts. 

R$~Ri 50,000 ohms 

The wave-forms of the potentials on the anodes and grids of the 
two valves of a symmetrical multivibrator are shown in Fig. 14. A 
symmetrical multivibrator is one which has identical valves, similarly 
biased, and in which all the components and wiring appertaining to 
one valve are precisely duplicated for the other. It will be seen that 
tne duration of the period between one change-over and the next is 



3° 



TIME BASES 



TYPES OF TIME BASE 



3* 



determined by the rate of rise of potential on the grid of that valve 
which has zero anode current at the time. If, therefore, the two coupling 
time constants have different values, the time periods of the two parts 
of each cycle will differ, i.e. the time during which one grid is positive 
will differ from that during which it is negative. The same results may 
be obtained by applying a small negative bias to one of the valves. 

Examination of Fig. 14 will show that the anode potential wave-form 
of each of the valves contains a negative "pip" which occurs when the 



:_r 



r 



r 



r 



r 

r 




Anode of V 2 



Anode of V z 



Qr/d of V l 



Grid of V z 



Fig. 14. Potential wave-forms for the multivibrator 

grid of the same valve is driven positive. The grid is made momentarily 
positive because the charge upon the condenser which feeds it has had 
time partially to leak away during the preceding period and, hence, it 
acquires a new charge by grid current. This necessarily means that, 
as the grid goes positive, the anode potential is reduced by more than 
the normal amount while the condenser recharges. This is a necessary 
part of the operation of the circuit so that if a more perfectly square 
wave-form is required at the output terminals some modification must 
be introduced in order to achieve the desired result. 

There is also a rounded corner which appears when the anode 
potential rises again. This is brought about by the stray anode to 
cathode capacitance. Increasing the value of the anode load resistance 
will tend to reduce the negative pip, but will make the corner more 
round than before. 

When a reasonably square wave-form is required, one may feed the 
output of the multivibrator into an amplifier which is overloaded on 



both the positive and negative swings so as to remove the irregularities. 
The principles described in Appendix VI may be used for this purpose. 

Another method which may be adopted is to employ pentode valves 
for the multivibrator circuit, but arranged with the screen of each valve 
connected to the grid of the other and with the anodes supplied through 
independent load resistances. The anode wave-forms of Fig. 14 then 
appear at the screens and by suitable choice of values for the anode and 
screen load resistances one may achieve a satisfactory wave-form at 
the anodes. This scheme is sometimes adopted when it is desired to 
switch the output of the multivibrator. For this purpose, a switching 
potential may be applied to the suppressor grids (see also page 354). 

As alternatives to the above methods, diode or other forms of limiters 
may be employed to limit the excursions of potential in the output 
circuit. 

The repetition frequency of the multivibrator may be controlled by 
arranging the circuit as follows: 

(a) The condensers C 1 and C 2 may be arranged in the form of two 
equal banks of condensers connected to a two-pole multi-contact 
switch. Each bank may consist of a number of units whose 
capacitance is increased by a factor of, say, 10 for successive 
contacts on the switch. 

(b) The resistances R x and R 2 may consist of two equal ganged 
potentiometers whose minimum and maximum values are in the 
ratio of, say, 1 : 11 in order to provide adequate coverage over 
the condenser ranges. This may be arranged by connecting fixed 
resistances in series with each element of an ordinary ganged 
potentiometer. 

Although the above method is often employed with satisfactory 
results, it suffers from the practical defect that ordinary ganged potentio- 
meters are not precisely matched and therefore the mark-space ratio 
is apt to vary appreciably as the potentiometers are rotated. An alter- 
native method, which permits the use of fixed resistances, so that the 
mark-space ratio may be maintained constant, is shown in Fig. 15, In 
this case, the grid leak resistances R x and R 2 are together connected in 
series with a variable bias potential which is common to both. It will be 
s ^en that the bias is obtained by means of a potentiometer connected 
across the supply rails. This provides a control of the repetition 
frequency over a range of the order of 10 : i, condenser switching 

mg employed as explained above if a wider range is desired. With 



this 



arrangement it is advisable to employ resistance values of the 



°rder of 1 or 2 megohms for R x and R 2 so as to avoid excessive 
grid current. 



32 



TIME BASES 



TYPES OF TIME BASE 



33 



Similar methods may be employed if it is desired to vary the mark- 
space ratio. For example, R x may be connected to the slider of the 
bias potentiometer while R 2 is taken to earth or to another separately 
adjustable potentiometer. The connection of the grid leak resistances 
to a high positive potential* results in stabilization of the period during 
which the valves remain in a non-conducting condition. 

The period of operation of a multivibrator may be calculated from 
equations given by Vecchiacchi. t Using T to indicate the complete 



OHT.+ 




Fig. 15. Modification of the multivibrator for change of repetition 
frequency without change of the mark-space ratio 

time period, * represents the time during which V 2 is cut off and T—t is 
the time during which V 1 is cut off. Then, with no applied bias (Fig. 13), 
E-V 0l 



t^C^ l0g e 



cZ 



and T-t^C 2 R X \og e E ~°* seconds, 



when R 2 > R a for V 1 and R x > R a for V 2 where R a is the internal 
impedance of the valves. 

In these equations, E is the supply potential, V ol the anode-cathode 
volts for the valve V 1 with supply volts E and with the intervalve 
coupling condensers removed, V oi is the same, but for V 2 , V cl , the 
static cut-off volts for a supply potential of E across the valve V x and 

V e2 is tne same f° r y * 

In order to make these measurements, it is, of course, necessary to 
provide a bias supply potential which is capable of adjustment between 
zero and the value necessary to cut off the anode current. Care should 
be taken to see that the anode supply potential remains, during the 
measurement, at the same value as obtains in the normal operating 

* E. L. C. White: British Patent 515768. 

t F. Vecchiacchi: "Meccanismo di funzionamento e frequenza del multivibratorc 
("The Mechanism of operation and the frequency of the multivibrator"). Alta 
Frequenza, 1940, 9, p. 745 ; G. W. O. Howe : " The Equivalent Circuit of the Multi- 
vibrator," Wireless Engineer, IQ45, 22, p. S7»i F. C. Williams: Introduction to 
Circuit Techniques for Radiolocation," jfourn. I.E.E., 1940, 93 < Part HlA,p, 2»9- 



conditions. It is advisable when the cut-off potential of one valve is 
being measured to set the other at zero grid volts. 

The complete period T is, of course, equal to the sum of the results 
of the above equations, i.e. T=t+{T—t). 

These equations follow simply from the consideration that when a 
condenser is charged or discharged through a resistance, the potential 
across the resistance decays exponentially. Thus, if the potential has 
a value A at a time t x and a value B at time t 2 , then the time interval 
between t x and t 2 is given by 

t 2 —t 1 =CR log e -- seconds. 
B 

This expression also applies when a fixed potential is present in series 

with the condenser and resistance. 

In the case of the multivibrator shown in Fig. 13, the initial potential 
across, say, R 2 when C x commences to discharge (corresponding to A) 
is equal to the initial potential E across the condenser minus the anode 
potential V ai across Vj. The final value of the potential across R % 
when V 2 is about to pass anode current once more, is equal to the 
cut-off bias potential V c2 . 

When a positive bias potential V b is inserted in series with the 
resistances, as in Fig. 15, the periods of the multivibrator are given by: 



t=C x R 2 log, * JL*±Z b seconds, 



and 



T-t = C i R 1 log, 



v c2 +v b 

E-V^+V, 



02.T _ b seconds. 



v. t +v k 

If it is desired to construct a multivibrator for use at a fixed frequency, 
it is a distinct advantage to make V h as high as possible, e.g. equal to 
E, in order to reduce the effect of variations in V ov V Q2 , V cl and V c2 
since the anode current in each valve then commences to flow while the 
appropriate condenser is discharging at a more rapid rate. If these 
potential variations are negligible in comparison with E, the periods 
then approximate to: ^^ ^ % % ^ A% 

and T— i=C t R 1 log e 2 seconds. 

Vccchiacchi's equations apply only to multivibrators having a low 
repetition frequency since they take no account of stray capacitance. 
Where the repetition frequency exceeds, say, 1,000 cycles per second, 
the following more precise equations should be employed : 

*=* 2 (C 1+ C 6 )log, (S-^ + ^i 



■ seconds, 



and 



7-*=*. fC, + CV) log, { g - y -+ |r *> C « seconds, 



34 



TIME BASES 




where C 5 and C a are the stray capacitances as shown dotted in Fig. 13. 
The above equations should not be used for extremely high repetition 
frequencies except as an approximation, since, in this case, other factors, 
including the effect of the stray capacitances C 3 and C 4 in Fig. 13, have 
to be taken into account. When the multivibrator is to operate at a 
high repetition frequency, or when it is desired to increase the slope 
of the wave-front, the anode load resistances should be reduced in value 
and the anode current increased. 

When the positive bias potential V b is applied to the circuit, the 
discharge time constant, i.e. the upward sloping part of the two lower 
curves of Fig. 14, tends to rise to a positive potential V b instead of to 
zero volts. As a result, the useful part of the curve is more steep* (see 
page 76) and the potential crosses the grid base of the valve more 
rapidly. This results in greater precision of operation and reduces any 
tendency to jitter. 

Jitter is a word used to describe certain erratic motions of the time 
base trace or of a wave- form under examination, which result from 
variations in the supply potentials or from stray pick-up. In general, 
jitter is the result of small variations in the supposed constant potentials 
appearing at the electrodes of one or more valves. For example, it 
will be obvious that a small potential appearing at the grid, if it varies 
at a rate which is not precisely the same as that of the synchronizing 
signal, can alter the time at which the grid acquires a given potential 
and, hence, can cause variations in the repetition intervals. 

The application of a succession of synchronizing signals to a multi- 
vibrator results in a change of repetition frequency of the latter. Usually 
this frequency increases and, when synchronization is achieved, the 
repetition frequency of the multivibrator bears an integral relation to 
that of the synchronizing source. Therefore, / s = nf M or f M = nf s where 
f s and f M are the repetition frequencies of the synchronizing source 
and the multivibrator respectively and n is an integer. Synchronization 
is more effectively achieved when f s = n/ jlf than in the opposite case. 
In the case when / s = nf Mt as the amplitude of the synchronizing 
signal is increased, the repetition frequency of the multivibrator also 
increases and n jumps to a smaller integer. 

The synchronizing signal may be injected in different ways and, as 
a result, the multivibrator will have a tendency to synchronize either 
at repetition frequencies which make n equal to an odd or an even 
integer or to have no discrimination. The following circuits, Figs. 16, 
17 and 18, show the method to be adopted for any desired arrangement. 
Obviously, if method (A) or (B) can be employed, any tendency for 

• Feinberg has also investigated the multivibrator. See R. Feinberg: "On the per- 
formance of the Push-Pull Relaxation Oscillator (Multivibrator)," Phil. Mag., 1948, 
39, p. 368: "Symmetrical Multivibrators," Wireless Engineer, 1949, 26, p. 153. 



TYPES OF TIME BASE 



35 



the multivibrator to synchronize at an undesired repetition rate is less 
likely to occur and more readily detected since the change will be twice 
as great as with method (C), It must be appreciated that when using 
methods (A) and (B) it is possible to synchronize the multivibrator 
with odd and even integer values respectively, but the tendency is greatly 
reduced. 
When pentodes are used for the multivibrator and the inter-valve 




■OH.T.+ 



OMI- 

Control 
"° Potential 
(in phase) 




GH.T.+ 



-OH.T- 



Fig. 16. Method A— 
even integers 




6 Control o 
Potential (in anti -phase) 

Fig. 17. Method B— 
odd integers 

Or/.T.-t- 



OH.T.- 

Control 

Potential 
(to one grid only) 

Fig. 18. Method C — any integer 

Methods of connecting a multivibrator for synchronization on even and odd 

integers 

couplings are arranged between the screens and the grids, the control 
potential may conveniently be applied to one or both of the suppressor 
gnds and the output may be taken from one or both of the anodes. 
I his arrangement has the advantage that the multivibrator is unlikely 

* ee d back a signal into the synchronizing circuit. 

hurrah* has described a method of controlling the frequency of a 
multivibrator. In Currah's circuit, variable slope pentodes (EF50) are 

sed for the multivibrator and triodes are placed across the pentode grid- 

c athode circuits so that the triode anodes are connected to the pentode 

* L. E. Currah: British Patent 599080. 



3« 



TIME BASES 



TYPES OF TIME BASE 



37 



cathodes and the triode cathodes to the pentode grids. The application 
of steady potentials across the triode grid -cathode circuits alters the 
internal impedance of the triodes and so varies the effective values of 
the grid leak resistances. In order to prevent fluctuations, the steady 
potentials should be supplied from ripple-free sources of low internal 
impedance. 

Friihauf's Time Base 

A circuit, due to Friihauf,* is shown in Fig. 19. This is a sym- 
metrical time base circuit in which one valve charges a condenser and 
the other valve discharges it. Resistances in the charging and dis- 
charging circuits are utilized to apply bias to the opposite valve so that 




Fig. 19. Friihauf 's time base 

only one valve functions at a time. These resistances carry the anode 
currents of the valves and hence they apply bias to one valve while 
removing it from the other. Thus, Ril a i must be sufficient to cut 
off anode current in V 2 . Similarly for R 2 I a & which must cut off F\ 
anode current. 

A modernized version of Friihauf's circuit is shown in Fig. 20 and 
the wave-forms associated therewith are shown in Fig, 21. This circuit 
is precisely the same as Fig. 19, except that one battery has been 
omitted and the other is replaced by an H.T. supply. In order to permit 
one side of C x to be connected to the negative supply rail and to earth, 
the resistance R t is removed from the cathode of V 2 and placed in the 
anode circuit of the valve V v Since R x (in Fig. 20) controls the anode 
current flowing in V 2 it is necessary to provide this control in another 
way. For this purpose, the time constant network C 2 R 3 is employed. 
Fig. 22 shows this arrangement. The operation of the circuit of Fig. 22 
is as follows. While the valve V 2 conducts, the condenser C 2 acquires 

* G. Friihauf: Arckiv Jiir Elektrotechnik, 1929, 21, p. 471. 



a charge, the upper plate becoming more positive. The time constant 
which determines the rate of charge is C^i^-fi?,), where R al is the 
internal impedance of the valve V x . The charging current, by passing 
■OH.T.+ 





off.Z- 



Fig, 20. Modification to Friihauf's 

circuit to avoid the use of 

separate H.T. supplies 



; Waye-form at 
! i Point A 



Wave -form at 
Point B 



Fig. 21. 

Wave-forms relating to 

Fig. 20 




VH.T.+ 



oQut 



through R t , applies a negative potential to the grid of V % via the net 
work C 2 R 3 . The time constant C 2 R 3 should 
be large compared with C x {R aX -YR\) or 
C\(R a2 -\-R 2 ), whichever is the greater, and 
R % should be large compared with R v 

When the potential across C x becomes 
sufficiently great to permit a small current 
to flow through the valve V 2 via R 2 > a poten- 
tial appears across R 2 which drives the grid 
of V x negative with respect to its cathode. 
The anode current in V v is thereupon re- 
duced in amplitude, the grid of V 2 is driven 
positive and C x is discharged via R 2 and 
V 2 with a time-constant C l (R a2 - ] fR 2 ), The 
change from the charging to the discharging 
condition is extremely rapid if the gain 
round the loop exceeds unity. 

These circuits produce a w r ave-form which 
is substantially triangular. 

Puckle's Time Base 

In 1933 the author* developed a time base which employs a variation 

•O. S. Puckle: British Patent 419298; U.S.A. Patent 2114938; "A Time Base 
employing Hard Valves," Journ. Television Society, 1936, 2, p. 147. 



oH.T.- 



FiG. 22. Modification of the 
circuit of Fig. 20 to permit 
the condenser C t and one of 
the output terminals to be 
connected to earth 



38 



TIME BASES 



TYPES OF TIME BASE 



39 



of the multivibrator as a condenser discharging medium. This raises the 
maximum repetition frequency, as compared with that obtainable 
from a thyratron time base, from about 40 kc/s up to a maximum of 
about 1 Mc/s. In this circuit (Fig. 23) a triode V 2 is connected across 
the condenser C 1 so that the cathode is free to make a voltage excursion, 
the anode potential remaining almost fixed. A pentode F 3 has a high 
value of anode load resistance and resides at zero grid potential during 
the charging regime. Since the anode of V z is directly connected to 
the grid of F 2 , the latter valve has its grid many volts (about 50 volts 
less than the H.T. supply potential) negative with respect to the cathode 
at the commencement of the charging period. As the condenser C x 
charges, it carries the cathode of V 2 progressively more negative. When 



OH.T.+ 




I — o Synch. 



Fig. 23. 

Puckle's time 

base 



oH.T.- 



the charge on the condenser is great enough to cause anode current to 
flow in the triode a small resistance R 2 in the anode circuit (variable 
trigger resistance) passes a negative potential to the grid or suppressor 
of K 3 . This reduces the anode current in F 3 and drives the grid of V 2 
positive so that a large anode current flows which discharges the con- 
denser. In this way, provided the gain from the anode of V 2 via V 3 and 
back to the anode of V 2 is greater than unity, the two valves reproduce, 
by cumulative action, the threshold effect of the thyratron and, in fact, 
behave in a similar manner, but without some of the disadvantages. 
The following values for the components will be found to be suitable: 

R 2 2,000 ohms (max.) 

ft 3 0-5 megohm 

R 4 0-25 megohm 

R 5 i-o megohm 

C 3 o*i jiF. 

C 5 o-i t*F. 

V x screen potential 20-150 volts. 



By controlling the value of the resistance R 2 , in the anode of V 2 , the 
trigger effect may be adjusted and, in general, the resistance requires 
to be reduced in value as the time base repetition frequency is increased. 
It will be clear that an increase in the value of R 2 will result in a decrease 
in the rate of discharge of the condenser and will therefore reduce the 
fly-back velocity. It will also, however, increase the rapidity with which 
the change-over from the charge to the discharge condition takes place. 
The variable resistance i? 4 , in the anode of F 3 , functions as a control 
of amplitude of the sweep voltage. As the resistance is increased, the 
voltage on the grid of V 2 is reduced, since the anode current is not 
altered, and the condenser has to acquire a greater charge before the 
cathode potential of V 2 is driven sufficiently negative to cause anode 
current to flow. 

The suppressor grid, or control grid, of the valve V z (whichever is 
not connected to the anode of V 2 ) is connected to a potentiometer to 
which a synchronizing signal is supplied in order to keep the time 
base operating in step with the wave-form to be examined (see 
page 239). 

The bias resistance and by-pass condenser in the cathode circuit of 
the charging valve V 1 are inserted to provide a form of D.C. feed-back. 
This enables more constant operating conditions to prevail if, in the 
course of time, it becomes necessary to replace the valve by a new one. 
This resistance is quite small and should be only sufficient to bias the 
grid to about — 1 volt. The anode current of V x is, of course, controlled 
by variation of the screen potential by adjustment of the potentiometer 
/?! in Fig. 23. Alternatively, the screen potential may remain constant 
and the anode current varied by alteration of the value of the cathode 
bias resistance. 

Although the fact was not appreciated when the circuit was developed, 
the valve V 2 may be regarded (during the discharge period) as a cathode 
follower which has a capacitive cathode load (C 1 in parallel with F x ). 
When F 8 is cut off the grid of V % is carried positive, and the cathode 
potential follows that of the grid. 

There is a maximum rate at which the condenser C x can be dis- 
charged ; the limit is due to the presence of the stray capacitances C 2 
and C 4) as shown in Fig. 24. The capacitance C 3 has little effect 
because the potential between the grid and cathode of V 2 remains small 
during the fly-back, and therefore no appreciable current flows into it. 
I he maximum rate at which the grid can go positive, even if C 1 is 
made to have a value of zero, is determined by the time constant 
"4(^2 + ^4). Thus, if R 4 is 200,000 ohms and C 2 -\-C^ is 20 micro - 
microfarads the time constant T is equal to 4 microseconds. With this 
value for T it is therefore useless to make C x (R a +R 2 ) much less than 



4 o 



TIME BASES 



TYPES OF TIME BASE 



4i 



4 microseconds, where R a is the average value of the anode-cathode 
impedance of V 2 . 

If we assume that R a is in the region of 2,000 ohms, we find that 
the value of C l which makes the two time constants 7? 4 (C 2 + C' 4 ) and 
C,(7^ a +^ 2 ) equal is 



Ci= (C 2 +C 4 )/? 4 = 4 x 



10 



-r, 



= o-ooi fiF. 



R a +R 2 4r° 00 

At first sight it would appear as though, with values of C 1 which 
are less than that given above, the cathode potential will rise faster 

-f —f 




Fig. 24. Effect of stray capacitance at the anode of V s in Fig. 23 

than that of the grid. If this happens, however, it produces a negative 
grid-cathode potential, and it is evident that the cathode must rise iti 
potential at the same rate as the grid. The potential of the cathode 
at every instant during the discharge may, of course, be slightly positive 
with respect to the grid. From the foregoing remarks it will be seen 
that this triode cathode follower is incapable of producing an extremely 
rapid fly-back, however small C 1 may be made, unless R A (C Z + C^ is 
also small. An improvement in fly-back velocity may, however, be 
obtained at the expense of an increased current taken from the power 
supply circuit if the valve F 3 is replaced by another valve of a type 
which has a larger anode current at zero grid volts, since this will permit 
the value of 7? 4 to be reduced. Obviously, if a valve of this type is to 
be employed, care must be taken to choose one which can have its anode 
current cut off by the potential available at the anode of V 2 . The value 



of R 2 cannot be increased to provide a greater potential for the grid of 
V$ without reducing the value of the discharge current from C x and 
thus increasing the effect which it is desired to reduce since this will 
increase the time constant C^R^ -fi? 2 ). 

An important practical detail is that a separate heater winding should 
be provided for V 2 in order to avoid leakage between heater and cathode. 
Furthermore, it is desirable that this winding should be screened from 
the H.T. and primary windings in order to avoid interference which 
would superimpose a mains frequency component on the time base 
wave-form, and thus produce a blurred or distorted oscillogram. 

Pulses of approximately square wave-form may be obtained from 
the anode load resistances of the triode V 2 and of the pentode V 3 of 
Fig. 23. These pulses can be used for the operation of other circuits, 
e.g. a black-out circuit to suppress the return trace. For this purpose, 
the anode of the triode V 2 may be connected via a condenser, which 
is capable of withstanding the operating potential of the cathode ray 
tube, to the grid of the tube. Alternatively the anode of V z may be 
similarly connected to the cathode of the tube. A modulating resistance 
or leak must, of course, be employed in either case. 

McMullan,* Keen and Den Hartog and Muller have described 
interesting modifications of the Puckle time base. McMullan's circuit 
is arranged to provide single-stroke operation and sweep expansion. 
Keen's circuit reduces the number of valves required from three to two. 

Keen*s Time Base 

The circuit of Keen's f time base is shown in Fig. 25. In this circuit 
a number of variations upon the circuit form of Fig. 23 will be noticed. 

(a) The discharge valve has become a high-slope tetrode which is 
capable of passing a high value of anode current. 

(b) The feed-back connection is taken from the screen of V % so that 
the fly-back velocity is increased by removal of some of the stray 
capacitance from the anode of V 2 , 

(c) The grid of the valve V x has its grid resistance connected to the 
positive instead of to the negative supply rail. 

{d) The use, if desired, of a short time constant for the network 

C 4 /? 9 . This differentiates the square-wave appearing at the screen 

of V 2 and provides a pulse at the grid of F,. As a result of this, 

C x commences to recharge before it is fully discharged so that 

the trace amplitude is reduced, but this is compensated by the 

fact that, since the discharge ceases before the anode potential of 

V 2 reaches the knee of the curve, the fly-back is linearized. 

D. McMullan: "A Single Sweep Time Base," Electronic Engineering, 1946, pp. 
21 and 47, 

T A. W. Keen: "Linear Saw- Tooth Generators," Wireless Engineer, 1948, 25, p. 210. 



4 2 



TIME BASES 



TYPES OF TIME RASE 



It will be seen that the screen of V 2 is coupled to the grid of the valve 
V x by means of the network C 4 i? 9 so that, when the condenser C x 
commences to discharge through V tr the potential of the screen of V 2 
falls and cuts off the anode current in V v The potential of the screen 
of V x rises and accelerates the discharge. 

The trace amplitude is adjusted by variation of the setting of the 
potentiometer R 4 , the fly-back time by adjustment of R$ and the 



o+ 300 V. 




■oOutput 



Synch. 



O-300V. 



Fig. 25. Keen's modification of Puckle's time base 

degree of synchronization by adjustment of the potentiometer jR 10 . 
Care should be taken to avoid the application of too large a synchroniz- 
ing potential since, otherwise, the value of the anode current in V t 
may be affected. 

Keen's time base, which is an important development, may also be 
used for single-stroke operation if the pentode V x is replaced by a 
triode-hexode, in which case the synchronizing potential may be applied 
to the grid of the triode portion, the triode anode being connected to 
the screen of the hexode portion of the valve. 

The resistances i? 2 and R 5 are intended to function as grid stoppers. 
The following components may be employed for the circuit: 



*t 


1,000 ohms 


C 


as required 


R 2 


330 ohms 


^2 


o-i y.F. 


R» 


47,000 ohms 


C 3 


10 ppF. 


i? 4 


o-i megohm 


c* 


o-ooi [iF. 


R & 


330 ohms 






R« 


2,000 ohms 


v, 


SP61 


*7 


4,700 ohms 


y% 


807. 


^8 


47,000 ohms 






*• 


047 megohm 






^10 


0-25 megohm 







43 



pen Hartog and Muller's Time Base 

It is sometimes required to see a fixed number of cycles on the screen 
of a cathode ray tube when the input frequency varies over a fairly 
wide range. A method of obtaining this result has been described by 
den Hartog and Muller* and the circuit arrangement is shown in Fig, 26. 
In Puckle's circuit, Fig. 23, the length of the trace varies as the syn- 
chronizing frequency is altered if the charging current remains constant. 
When using the circuit of Fig. 26, however, the valve V \ rectifies the 



O300V+ 




Synch 



Fig. 26. Den Hartog and Muller's modification of Puckle's time base 



trace potential and employs a portion of it to control, by a feed-back 
method, the value of the charging current. 

The valve V A is a cathode follower and two diodes provide a rectified 
potential across the condenser C e so that, as the trace length increases, 
the rectified potential is reduced. A portion of the potential across C 6 
Is applied to the grid of F 3 so that the charging current (anode current 
of P 8 ) is also reduced. 

a he meter in the anode circuit of the charging valve F 3 is now able 
to indicate the repetition frequency which is substantially proportional 
t0 the charging current. 

The following values may be employed for the components for a 
re qucncy range of 150 to 2,000 cycles per second: 

, H. den Hartog and F. A. Muller: "Oscilloscope Time Base Circuit," Wireless 
tnumeer, IO , 47p 2 4, p. 287. 



44 





TIME 


BASES 




*l 


50,000 ohms 


c\ 


0-005 v-F 


JKt} 


700,000 ohms 


c 2 


1600 [xF 


R* 


150,000 ohms 


c 3 


o-i (iF 


R* 


100,000 ohms 


Q 


o-oi |i.F 


R* 


2,000 ohms 


C 5 


1*0 ^F 


R* 


r 1,000 ohms 


c« 


0-5 fiF 


*7 


4,000 ohms 






#« 


50,000 ohms 


Vi 


EF6 


R* 


20,000 ohms 


v* 


EL3 


Km 


100,000 ohms 


v 3 


EF5 


Rn 


1,600 ohms 


v, 


EBC3. 


R u 


5 megohms 






^13 


3 megohms 






Ru 


10 megohms 






•#15 


250,000 ohms 







TYPES OF TIME BASE 



45 



Moss's Time Base 

Another circuit, not based on the Puckle time base, which achieves 
similar results has been described by Moss.* 

Moss's circuit provides extremely rigid locking, in fact, this is the 
principal feature of his arrangement. 

The idea of the inherent rigid locking of a time base with the wave- 
form under examination is not new. Nagy and Goddardf have described 
a method of accomplishing this extremely desirable effect but their 
method involves the use of a very special type of valve which incor- 
porates two surfaces having secondary emission ratios which differ in 
value. The necessity for this special valve is a serious objection and 
the advantage of Moss's method is that his circuit operates with normal 
types of valves. 

The basic principles from which this circuit has been designed may 
be described as follows. The wave-form to be examined is squared by 
the valve V \ (Fig. 27), a cathode follower (not shown) being inserted 
before the squaring valve to avoid any distortion of the wave-form as 
it appears upon the screen of the cathode ray tube. The squared signals 
are fed via the gate valve V 2 to the grid of a valve V A which reverses 
the phase of the signal before applying it to the discharger valve V 6 . 
The valve F 7 is a valve forming part of the News am time base (see 
Fig. 99). 

The circuit functions in the following manner. The time base itself 
consists of the valves V Q and V 7 and their associated components. The 

* H. Moss: "Time Base Converter and Frequency Divider," Wireless Engineer, 
1945, 22, p. 368. 

f P. Nagy and M. J. Goddard: "The Signal Convertor," Wireless Engineer, 1043, 
20, p. 273; P. Nagy: "The Signal Convertor and its application to Television," Joitrn. 
Television Society, 1944, 4, p. 26. 



discharger valve is switched on and off by the squared form of the 
signal which is to be examined. However, circuits of this type are very 
old and the special feature of Moss's circuit is an arrangement which 
permits the discharge valve V 6 to be switched on once in every n cycles 
of the input wave-form in order that n cycles may be made visible on 
the screen of the cathode ray tube. In order to bring this idea to 
fruition, a gate or strobe valve V 2 is inserted in series with the squaring 
valve V-i and the output is taken from the anode of V 2 . Unless both 
V-i and V 2 are conducting at the same instant, no output signal is fed 

Jts Us R 7 A 8 *f *' 



+300V 

o- 



T TTTTT, 

V3C4.Cs Cff C? Cg 




+1000 V 
-o 



Pig. 27. Moss's time base 

to the grid of F 4 . The valves V 1 and V 2 may, if desired, be replaced 
by a single valve of the type (VR116) in which either the grid or the 
suppressor may be employed to cut off anode current. 

Let us assume, for the moment, that F 2 is conducting and that the 
signal to be examined is applied, via a cathode follower, to the input 
terminals of the circuit shown in Fig. 27: a squared form of the input 
signal will appear at the grid of F 6 in phase with the signal at the grid 
of Fj. Let us suppose that this signal is positive so that anode current 
commences to flow in V e . The condenser C 14 commences to discharge 
and the grid of valve V 7 goes negative. After some time, due to the 
delay in the network R 5 R 9 , C 3 C 6 , the potential at the grid of V 2 
also falls until V 2 anode current is cut off. No further pulses can be 
fed to the grid of V 6 while this condition exists. Before V 2 is cut off, 
the condenser C 14 will be discharged, and upon cut off of V 2 it com- 
mences to charge once more until eventually the grid potential of V 2 is 
brought to a point within the grid base of the valve. In this condition, 
when the grid of V 1 is again driven positive another positive signal 



46 



TIME BASES 



appears at the grid of V s and the cycle of operations recommences. 
The following facts will be appreciated: 

(a) V 1 must be cut off during the negative portions of an input 
signal. This is brought about by the use of a large input signal 
— not less than about 20 volts peak to peak. Too small a signal 
also results in a slow fly-back. 

(b) V 2 must have its grid potential brought within the grid base in 
time to transmit a square wave via V 1 and its potential must not 
fall until after the condenser C 14 has been discharged. The delay 
due to the line must therefore exceed the duration of the square 
wave by, say, 50 per cent. 

(c) V 2 must be normally cut off—this potential is adjusted by 
variation of the preset potentiometer R 1&i which also affects the 
number of cycles appearing on the screen. The grid of V 2 is 
normally negative with respect to the cathode of V l and the 
diode V z limits its excursion in the positive direction to that of 
the potential of the cathode of V v This, however, permits only 
a very small potential to appear across the anode of V 1 and 
improved results would be obtained if the cathode of V z were 
connected instead to a suitable point on the resistance chain i? 15> 
R 16 , i? 17 . The most suitable point is probably at the junction 
of R l5 and R u . 

{d) The potentiometer R l9 is preset at a value which is just more 
than sufficient to cut off anode current in V 6 in the absence of 
any input signal. 

(e) The resistance i? 20 controls the number of cycles of the wave- 
form which appear on the screen of the oscillograph. 

It should be noted that the component values which are given below 
apply to a circuit for examining a 50-cycle input wave-form: 



R, 


2 megohms 


c % 


1-0 [aF. 


R* 


100,000 ohms 


L»2 


O OI [lF. 


*3 


150,000 ohms 


^"3 — ^ 8 


0*002 nF. each 


#4 


50,000 ohms 


c 9 


0-5 (xF. 


«*-*« 


80,000 ohms each 


^10 


1 uT. 


-^10 


800,000 ohms 


c u 


01 fjtF. 


fill 


10,000 ohms 


^12 


4-0 -aF. 


R\z 


20,000 ohms 


^13 


32-0 hF. 


&13 


500 ohms 


t'14 


0-35 |*F. 


/? 14 


1 megohm 


c„ 


2-0 U.F. 


«H 


30,000 ohms 






^16 


100,000 ohms 


Vi 


41 MTL 


R 17 


30,000 ohms 


v 2 


41 MTL 



TYPES OF TIME BASF. 



47 



Rm 


30,000 ohms 


v, 


VR78 


■^18 


100,000 ohms 


v, 


41 MTL 


^2 


2 megohms 


Vs 


VR78 


«£1 


70,000 ohms 


v* 


42 SPT 


R22 


70,000 ohms 


v 7 


41 MTL 



Fig. 28 shows the wave-forms appearing at different parts of the 
circuit and a study of the figure will make more clear the method of 
operation of the circuit. 

-LrirTrrru" ■*> 

1; i' 




1 1 11 

Fig. 28. Idealized wave-forms pertaining to the circuit of Fig. 27 

(a) Square pulse derived from the input signal. 

(b) Wave-form applied to the grid of V 3 . 

Note the phase retardation due to the delay network. 

(c) Positive output pulse from the anode of V 4 . 
id) Time base wave-form at the output terminal. 

Fig. 29* shows a photograph of a square wave-form as obtained with 
this oscillograph. It should be noted that although the photographic 
exposure time was 20 seconds, no variation in the fly-back position is 
visible. This is the result of the extremely rigid locking which is 
inherent in the circuit, and this, in turn, is due to the absence of 
metastable conditions. 

Potter's Time Base 

Potter f has developed a time base which employs a cathode-coupled 
multivibrator. The circuit is shown in Fig. 30. The charging circuit 

* Facing p. S o. t J- L. Potter: " Sweep Circuit," Proc. I.R.E., 1938, 26, p. 713. 



4 8 



TIME BASKS 



TYPES OF TIME BASE 



49 



consists of the resistance R& and capacitance C 3 while V x and V 2 form 
a trigger circuit for discharging C 3 through V 2 . In order to discuss 
the operation of the circuit it is necessary to decide upon a starting point 
in the cycle of operation. Commencing, therefore, at the point at which 
anode current in the valve V x begins to increase, the anode current in 
V 2 will cease due to the negative potential appearing at the grid of V 2 
via C 2 , and the anode potential of V 2 will rise at a rate determined by 
the time constant RtC 3 and the potential of the supply rail. During 
the relatively long period while C 3 is charging, C 2 discharges through 
V lr R 3 and i? 5 , and eventually the potential at the grid of V % will have 
risen sufficiently to permit a small flow of anode current through V 2 ; 



0//.7> 



Out 




Control 
Potential 



Tr i } B±c ? 

■-1 - -i — 1 — 1 i — oMt- 



Fig. 30. Potter's time base 

the current through 7? 3 then increases, which increases the bias on V t 
and reduces its anode current. This raises the potential at the grid of 
V 2 and, provided the gain round the circuit including V 1 and the 
grid and cathode only of V 2 is greater than unity, the resultant cumu- 
lative action cuts off the anode current in V ly causing the anode current 
in V 2 to rise to a maximum which thus discharges C 3 once more. 
Reverting to the beginning again, as the anode current in V x increases, 
the cathode potential rises, but not very greatly, because the cathode 
current of V % is, at the same time, decreased. When, later, C 3 discharges, 
C 2 charges via R 2 , R z and the grid-cathode path through F 2 . As C 2 
becomes charged to an increasing potential by grid current in F 2 , the 
common cathode potential falls as the currents through C 2 and C 3 fall. 
Since, apart from the current in R 3 and the presence of a synchronizing 
signal, the grid potential of V x is constant, F x can be cut off only by 
an increase in the current through i? 3 . Therefore, the valve V 2 must 
be capable of drawing more current from C 3 than V x can pass via R 2 
when it is self-biased by R 3t since, otherwise, K x cannot be cut off. 
The fly-back, or the cessation of the discharge of C 3 , is terminated 
when V 1 commences to conduct once more, without regard to the value 
of the potential to which C 3 has actually been discharged. 



The wave-forms generated in various parts of the circuit are shown 
in Fig. 31. 

In order to obtain a reasonably rapid fly-back, R 2 and R B should be 
of low resistance. Potter recommends values of about o-i to 0-5 megohm 
and 200 to 500 ohms respectively. R A and R b> which should be variable, 
should be of the order of 1 megohm each and may conveniently be 
ganged and the H.T. supply potential should be about 500 volts. The 
capacitance of C 3 should be as small as possible in order to reduce 



(a) Anode Potential 

of y\ 





. (b) Cathode Potential 
of V 2 and V 2 



(c) Grid Potential 
ofV 2 



(d) Anode Potential 
ofV 3 



Fig. 31. Wave-forms pertaining to the circuit of Fig. 30 

fly-back time and this condenser should be permitted to charge only 
to a small percentage of the rail potential to avoid serious non-linearity. 
For this reason, Potter's time base generally requires the use of an 
amplifying stage V s when it is used in conjunction with a cathode 
ray tube. 

The condensers C 2 and C 3 must be changed together when changing 
from one frequency range to another. 

Because short positive pulses are obtainable at the cathodes, the 
circuit is very useful for incorporation in a multi-stage frequency 
divider. The cathode output of each stage is, for this purpose, applied, 
preferably via a cathode follower, to the grid of V l of the next stage 
which may be adjusted to have an operating period of, say, three to five 
times that of the first stage. The second stage may be arranged to 
control further stages. Twin triode valves may be used for Potter's 
circuit. 
6 



5° 



TIME BASES 



T Anode - Grid 
1 Volts Volts 



INVESTIGATION INTO THE PERFORMANCE OF A 

TIME BASE 

The investigation of the processes taking place during the operation 
of a time base is one which is best carried out with the aid of a cyclo- I 
gram. An example, given in Fig. 32, shows the behaviour of Puckle's 
circuit when the repetition frequency is about 1,000 cycles per second. 
In the upper diagram the excursions of the cathode of the discharge 
triode V % are shown horizontally and those of the grid vertically. 
Starting from the bottom left-hand corner it will be seen that the grid- * 
cathode potential of the triode remains negative, due to the presence | 
of anode current in the pentode V z% while the triode cathode becomes 
increasingly negative as the condenser charges. When the potential of I 
the triode cathode approaches that of the grid, anode current flows and 

the grid potential rises rapidly, 
as shown by the vertical por- 
tion at the right-hand side of 
the cyclogram. This process 
is modified when the rise of 
grid potential is limited by the 
flow of grid current through 
i? 4 . The potentials of the grid 
and cathode then move to- 
gether as indicated by the 
straight line at 45 °. (Actually 
the slope only approximates to 
45 even when the cyclogram 
is redrawn to allow for the 
unequal sensitivities of the X 
and Y plates of the cathode 
ray tube on which the cyclo- 
gram is being observed.) The 
cathode follower process ceases 
abruptly at the top corner of 
the cyclogram where the grid of F 2 is driven violently negative. When 
the condenser discharge current falls in value towards the end of the 
discharge period, the negative potential on the suppressor grid of F 8 
is so far reduced as to allow a small amount of anode current to flow. 
The pentode V z then drives the triode. grid negative, thus stopping the 
discharge, and the cycle repeats itself. The lower diagram shows the 
excursion of the triode anode in a vertical direction and of the triode 
cathode in a horizontal direction. 

Note that the cyclogram will vary somewhat when the repetition 




v 2 v 3 

Anode - Suppressor 
Volts Volts 



Vj Anode Volts 
= V 2 Cathode Volts 
Condenser charging P* 

FlG. 32. Cyclogram of Puckle's time base. 

Frequency approximately 1,000 repetitions 

per second 



+ -*- 




Fig. 2») 








Fig. 33. Cyclogram. 

Frequency tipptoximnlely one 

repetition per second 



i""in. 34. Cyclogram. 

Frequency approximately 250,000 

repetitions per second 



[To face page 50 



TYPES OF TIME BASE 



5i 



frequency is changed. Fig. 33 shows the cyclogram for approximately 
one repetition per second, and Fig. 34 for approximately 250,000 
repetitions per second. At the higher frequencies the stray capacitance 
between the grid of F 2 and the positive supply rail slows down the 
rate of rise of the grid potential, and therefore impairs the rate of 
discharge of C v 

Although time is not one of the co-ordinates of a cyclogram it is 
implicit therein and can be roughly estimated. The faster the spot is 
moving the fainter is the line produced. If a more precise time measure- 
ment is required, a time-marker pulse may be applied to the grid or 
cathode of the cathode ray tube so as periodically to quench the beam 
and so to provide dark patches having equal time separation. These 
dark patches will then be crowded or well spaced according to whether 
the spot moves slowly or fast. 

Further examples of performance investigation by means of cyclo- 
grams are given in Appendix III, p. 319. 



TRIGGERED TIME BASES 



S3 






CHAPTER IV 

TRIGGERED, SINGLE-STROKE OR EXTERNALLY 
OPERATED TIME BASES 

Many time bases are externally triggered so far as the charging device 
is concerned, the switching device being self-initiated, but sometimes 
the reverse is the case, i.e. they perform only one cycle of operation on 
each occasion upon which a control signal of suitable type is applied 
to the appropriate electrode of one of the valves in the circuit. Such 
an arrangement is said to be triggered or externally operated. Since 
this type of circuit operates only once upon the receipt of each triggering 
pulse, it is clear that it is also a single-stroke time base. Any circuit 
which is to function as a triggered time base must have a stable condition 
in which it can rest w r hen no triggering pulses arrive. 

It is appropriate at this point to consider the difference between a 
triggered time base and a trigger circuit. Every time base has two 
regimes, a charging regime and a discharging regime. Normally, one 
of these regimes is unstable and sometimes both are unstable, i.e. the 
circuit cannot remain in the charged or the uncharged condition, but 
must automatically commence to reach the opposite condition. A trigger 
circuit is one which produces this instability. When one or both of the 
regimes is stable an external triggering, or control, pulse or other form 
of signal is required to initiate each cycle of operations and a time base 
circuit which operates in this w r ay is known as a triggered time base. It 
will thus be clear that a trigger circuit is only a part of a time base. 

Sometimes the triggering pulse is applied a short time before the 
wave-form to be examined is applied to the cathode ray tube in order 
that the commencement of the wave- form may be visible. A triggering 
pulse which temporarily precedes the wave-form to be examined can 
sometimes be obtained from the circuit which produces the wave-form. 
When this condition does not exist, the wave-form can be applied via 
a buffer amplifying stage followed by a differentiating circuit (see Appen- 
dix IV) to produce a triggering pulse, while the wave-form is also fed, 
via a correctly terminated delay line (see Appendix VI), in order to delay 
its arrival at the Y plates of the cathode ray tube by a sufficient time 
to allow the trace to commence before the arrival of the wave-form. 
The duration of the trace is adjusted to be sufficient to permit the 
required part of the wave-form to appear. Thus, if the time base is 
produced by the charging of a condenser, the latter should remain 
discharged pending the arrival of the triggering pulse since, otherwise, 
valuable time is lost in discharging it. 

52 



Nearly all time bases which were developed prior to 1939 were so 
arranged that the synchronizing or triggering pulse operated to dis- 
charge the condenser, while the modern tendency is to initiate the 
charging of the condenser by the synchronizing signal. 

Some triggered time base circuits, instead of being entirely quiescent 
in the absence of a triggering pulse, continue to operate at an extremely 
slow repetition rate. 

When the time base circuit is required to function in connection 
with a cathode ray tube, it is usually necessary to arrange that the 
cathode ray beam shall be blacked out or deflected off the screen until 
the sweep commences in order to avoid "burning" the screen. This 
also serves to increase the apparent brightness of the resulting trace by 
removal of the very bright light which otherwise exists when the spot 
is stationary. In order to save time at the commencement of the trace, 
it is advisable to adopt the blacking-out procedure. 

Many of the time bases which are described in later chapters are 
triggered time bases but they are grouped with others because they 
have some other distinctive feature in common. For this reason, only 
one triggered time base will be described here. 

Smith's Triggered Time Base 

Smith* has developed a circuit which is a combination of Potter's 
time base shown in Fig. 30 and Schmitt's trigger circuit shown in 
Fig. 58 together with the addition of two limiting diodes. In Smith's 
time base, the lower end of the resistance R b of Fig. 30 is connected to 
the junction of the cathodes of the valves V l and F 2 , but he has added 
an additional resistance in the cathode circuit of V 2 only. This has the 
effect of making the circuit reside in a stable condition in the absence of 
a positive triggering pulse which is normally applied to the grid of V v 
The reason for this is explained on page 83. 

In addition, there are two potentiometers connected in series across 
the H.T. supply rails and the diodes are connected thereto in the 
following manner. One diode has its anode connected to the anode 
of V u which is here changed to a pentode, and its cathode is connected 
to the slider of the potentiometer which is nearer to the positive rail. 
The potential excursion of the anode of V t is thus prevented from 
exceeding, in the positive direction, the value determined by the setting 
of the slider. The second diode has its cathode connected to the anode 
of V 2 and its anode connected to the slider of the remaining potentio- 
meter so that the potential excursion of the anode of V 2 is limited, in a 
negative direction, by the setting of the slider to which it is connected. 

The diode connected to V % is intended to ensure that K 2 continues to 
•L. C. Smith; British Patent 604763. 



54 



TIME BASES 




conduct current and, hence, that V l is substantially non-conducting 
in the quiescent condition. 

This proves to be necessary because Smith had placed an additional 
bias resistance in the cathode circuit of V 2 . Without this resistance, 
the diodes become unnecessary. 

Positive pulses are applied to the control terminals, at irregular 
intervals if desired, and each pulse causes one sweep of the time base 
potential. Since the time base condenser C 3 is held discharged by K 2 , 
which normally conducts, no time is wasted in discharging the con- 
denser before a new sweep can begin. 



CHAPTER V 
TRIGGER CIRCUITS 

Although every self-operated saw-tooth time base incorporates a 
trigger device of one kind or another, this chapter will be confined to 
a few trigger circuits which appear to be especially closely related to 
one another. Some trigger circuits of a different type are discussed in 
Chapter IX. The circuits in this group are readily adaptable to a wide 
variety of problems, and they are frequently employed to provide a 
trigger action for bringing about the operation of a further circuit 
either automatically or upon the receipt of an initiating signal. The 
members of this family may be divided into three categories ; those in 
which the trigger action is obtained in a single valve by the judicious 
selection of operating conditions so that amplification occurs between 
one electrode and another, without the usual reversal of phase, those 
in which phase reversal occurs between each of two pairs of electrodes, 
usually with amplification in the case of one pair only, and those in 
which a back-coupled pair of valves is employed. 

The occurrence of a trigger action may be explained by demon- 
strating the existence of a negative resistance in the circuit, as indicated 
by a negative slope over a region of the voltage-current characteristic, 
obtained by direct measurement or by calculation.* Alternatively, it 
may be regarded as a consequence of positive feed-back in an amplifier. 
Thus, a two-valve amplifier provides amplification without phase 
reversal and, if the output is fed back to the input, the circuit will 
generally oscillate because the feed-back is positive. In a single-valve 
amplifier the output voltage is normally in phase opposition to the 
input so that back-coupling generally increases the stability of the 
circuit and does not tend to produce oscillation, unless a large phase 
change is introduced in the feed-back network. In order to construct 
a single- valve trigger circuit one may nevertheless employ the same 
principles as in a two-valve circuit, provided that the valve is used in 
such a way that amplification is obtained without phase reversal. 
Under certain conditions a pentode may be used for this purpose by 
applying a controlling potential to the suppressor grid and obtaining 
an amplified potential change at the screen grid. Feed-back from the 
screen to the suppressor then results in the production of oscillations 

<( • H. J. Reich: "Trigger Circuits," Electronics, August 1939, p. 14; E. W.Herold: 
'Negative Resistance and Devices for obtaining it." Proc. I.R.E., 1935, 23, p. 1200; 
P. lc Corbeiller: "The Non-linear Theory of the Maintenance of Oscillations," Journ. 
f-E.E., 1936, 79, p. 361 ; S. P. Chakravarti: "On the Nature of Negative Resistance 
and Negative Resistance Sections," Phil. Mag., 1940, 30, p. 294. 

55 






56 TIME BASES 

of various forms, which are comparable with the oscillations obtainable 
with the back-coupled two-valve amplifier. 

In most cases the trigger circuit is arranged to come into operation 
when the charge upon the time base condenser reaches a predetermined 
potential and to function in the opposite sense when the charge has 
been brought to another predetermined value. This usually means 
that the first trigger action is equivalent to the closing of a switch, and 
the second action is equivalent to opening the switch, but, to make 
the analogy more complete, the switch must be of the snap-action 
toggle type. 

These circuits have two unstable limiting conditions when they are 
employed in a self-running time base. The existence of two limiting 
conditions is basically due to the fact that a valve may be limited in 
two ways: by driving the control electrode sufficiently positive or 
sufficiently negative. The two limiting conditions are both unstable 
because the time base condenser is always either in the process of being 
charged or of being discharged. If the circuit is arranged so that a 
stable condition can be reached, the charge on the condenser becoming 
constant, it forms a single-stroke relaxation oscillator, and only one of 
the limiting conditions is unstable. Upon the arrival of the initiating 
signal the trigger circuit "flips" over from the stable limiting condition 
to the unstable limiting condition and " flops" back after a definite 
predetermined time interval. The term Flip-Flop is applied to circuits 
of this type, and they are generally employed to actuate single-stroke 
time bases and similar devices. Any relaxation oscillator may be over- 
biased so that only a single operation results from each initiating pulse. 
In the case of the flip-flop this is a "go and return" operation, but 
there is another type of circuit having two stable limiting conditions 
in which alternate initiating pulses flip and flop the circuit. This 
circuit will be described later. 

The Transitron Relaxation Oscillator 

One of the earliest forms of single- valve trigger circuit was described 
by van der Pol* in 1926. This circuit employs a tetrode in which the 
screen and grid have a resistance-capacitance coupling, the grid being 
fed from the high potential source via a high resistance, which forms 
part of the coupling network. Van der Pol postulates that it is essential 
to have an inductance, no matter how small, in series with the screen 
coupling circuit, as shown in Fig. 35, but later investigators appear to 
be of the opinion that this inductance is not necessary. 

The term "transitron" has recently come into use to denote any 

• B. van der Pol: "On Relaxation Oscillations," Phil. Mag., 1926, 2, p. 978; "The 
Non -linear Theory of Electric Oscillations," Proc. I.R.E., 1934, 22, p. 1051. 



TRIGGER CIRCUITS 



57 







-OH J* 



circuit which employs a single pentode valve in such a way that amplifi- 
cation is possible without phase reversal. The name was originally 
coined by Brunetti, who defined it as "a retarding-field negative- 
transconductance device." Generally, these circuits* are extremely 
versatile, and one may produce a sinusoidal, saw-tooth or square wave 
output from one oscillator by means of very simple switching. It is 
equally easy to convert the continuous oscillator into a flip-flop by 
changing the bias potential. When the circuit is arranged as a relaxation 
oscillator or as a flip-flop it is very valuable as a switching device, and 
has many applications to time bases and control circuits. 

The fundamental basis on which these circuits are constructed is 
that when a change of potential is applied to the suppressor grid it 
appears in amplified form, with- 
out phase reversal, on the screen 
grid. The reason for this is as 
follows: 

If the anode current of a 
pentode is interrupted, electrons 
continue to pass between the 
wires of the screen grid, but 
they are no longer collected by 
the anode and must therefore 
return towards the screen (pos- 
sibly passing back and forth several times before finally reaching 
the screen wires) and therefore the total cathode current in a pentode is 
constant provided the screen and control grid potentials remain un- 
altered. As a result of this the screen current is increased when the 
anode current is reduced. If, however, the screen is connected to a 
high-potential source of supply through a high resistance, the change 
in screen current will be much smaller than the change in anode 
current, but, on the other hand, there will be a considerable change in 
the screen potential. 

By varying the potential of the suppressor grid the anode current is 
controlled and this, in turn, controls the screen current and, hence, 
its potential, so that when the suppressor potential is made more nega- 
tive the screen potential becomes less positive. A pentode which is 
provided with a high resistance in series with the screen may therefore 
be used as an amplifier, the input being applied to the suppressor 
(suitably biased) and the output, in the same phase as the input, 
being taken from the screen. An amplifier of this type becomes an 

• C. Brunetti: "The Clarification of Average Negative Resistance with Extensions 
of its Use," Proc. I.R.E., i 9 37> 25, p. 1595 ; " The Transitron Oscillator, Proc. I.R.E.. 
1939, 27, p. 88; C. Brunetti and E. Weiss: "Theory and Application of Resistance 
Tuning," Proc. I.R.E., 1941, 29, p. 333. 



Fig. 35- 



Van der Pol's transition circuit 



58 



TIME BASES 



^ 



oscillator if a positive feed-back circuit is arranged between the 
input and output terminals; i.e. between suppressor and screen. For 
example, with a resistance-capacitance coupling between the suppressor 
and screen the device becomes a relaxation oscillator and is almost 
equivalent to a multivibrator. It is, however, essential that, in the 
absence of the coupling condenser, the amplification shall be greater 
than unity. If the standing bias on the suppressor is such that the 
amplification is less than unity, the device becomes a flip-flop. It is, 
of course, assumed that the amplification will exceed unity at some 
smaller value of bias potential. This application will be described later. 

The Transitron Sinusoidal Oscillator 

The transitron sinusoidal oscillator is shown in Fig. 36. It is included 
here for the sake of completeness although, of course, it is not a time 

-OMJ.+ 




oNX- 



oB/as-ZOV. 



Fig. 36. The transitron sinusoidal oscillator 

base. As in all oscillators, this circuit provides a negative resistance 
and it is only when the negative resistance is equal to or less than the 
(positive) resonant impedance of the oscillatory circuit that oscillation 
will be maintained, since only under the above conditions are the circuit 
losses made good and the oscillation prevented from decaying. The 
value of the negative resistance which appears in shunt across the 
oscillatory circuit is found as follows: 

Let y! denote the amplification factor (considered as positive) between 
suppressor and screen and R s the internal impedance between screen 
and cathode, both values being measured under the operating con- 
ditions with the anode circuit complete. Then the screen current 



t.~- 



Rs+Rs 

where E is the H.T. potential, e jS the applied suppressor potential and 
i? 3 the screen feed resistance. This equation is similar to that employed 
in calculating the anode current of a triode, which is 



TRIGGER CIRCUITS 



59 



*„= 



R 2 -\-R a 



where e gl is the grid potential, u. is the amplification factor between 
grid and anode (not the same as in equation (1)), R 2 the anode load and 
R a the internal anode-cathode impedance. The reason for the different 
sign in the two numerators is that in the former case no phase-reversal 
takes place. 

When the screen and suppressor are connected together by means of 
a condenser whose impedance is negligible at the frequency of oscilla- 
tion, the alternating potential on the suppressor is equal to the alter- 
nating potential on the screen, or 

e gZ =E-i s Rs ... (2) 

then, by eliminating e g8 from equations (1) and (2) 

. E-£(E-uRi) 
R a +R s 



t*=- 



g(i-V) 
a, (i -*')+*« 

Hence, it will be seen that the impedance offered to the H.T. supply 
rail by the screen circuit is 



E 



=R*+ 



R. 



1— ii 



=* 3 -r 



1 
R. 



R, 



The effect of the screen-suppressor coupling is, therefore, to place 

R 

a negative resistance equal to *, in parallel with the internal resis- 

tance R a . The impedance which shunts the tuned circuit LC therefore 
consists of R Zt R s and — * all in parallel. R 3 appears in parallel because 

the positive and negative supply rails are at the same potential in the 
A.C. sense due to the presence of the final condenser in the rectifier 
smoothing system which is of negligible impedance at the operating 
frequency. The value of the negative resistance may be controlled by 
inserting a variable bias resistance in the cathode circuit. This resistance 
will also control the purity of the output wave-form. 

When using an EF50 valve under the operating conditions shown 
below, R B is about 25,000 ohms and y. is of the order of 4, so that 
the negative resistance, after allowing for i? 3 (50,000 ohms) is about 
10,000 ohms. 



6o 



TIME BASES 






The existence of negative resistance is essential to any oscillator or 
trigger circuit, and the principles involved in the method of calculation 
described above are generally applicable. 

The output may be taken from the anode of the valve or from the 
screen or suppressor. In practice, however, it is advisable to draw the 
output from the anode since, in this case, the impedance of the load 
has less effect upon the operation of the circuit. 

Suitable values for the components of Fig. 36 are as follows: 



10,000 ohms 
50,000 ohms 

to provide the required frequency / 



R 2 

In 
C x 

C 2 say 0*005 \tP. (impedance to be negligible at frequency/) 
H.T. 350 volts. 

The Transitron Saw-Tooth Oscillator 

Fleming- Williams* and Reichf have independently developed almost 
identical time base circuits in which a single valve is used to charge or 
discharge a condenser connected between the anode and one of the 
supply rails. Fleming- Williams' version of the circuit is shown in 

Fig- 37- 

The mechanism of the complete circuit may be understood by 
commencing at the period when the suppressor is positive and the 
anode is taking current. The anode current is then considerably larger 
than the charging current through R 2l and therefore the condenser C x 
is discharged rapidly. As the anode voltage falls the anode current 
remains approximately constant until the knee of the characteristic is 
reached, when the value of the current commences to fall. This results 
in a rising screen current, a falling screen potential, and a falling sup- 
pressor potential. This, in turn, reduces the anode current still more 
and the process becomes increasingly rapid until the velocity is limited 
by the stray capacitances of the circuit. Thus, the anode current is 
suddenly cut off, while the suppressor is driven negative. This con- 
dition remains while the condenser C x is being charged through R z . 
As the condenser charges the anode potential increases and a point is 
reached at which the anode commences to draw current which would 
otherwise pass to the screen. The suppressor-grid potential therefore 
rises once more, since the suppressor and screen are coupled, until the 
anode suddenly takes a large current, thus discharging the condenser C x . 

* B. C. Fleming- Williams: "A Single-Valve Time Base Circuit," Wireless Engineer, 
1940, 17, p. 161. 

t.H. J. Reich: "Trigger Circuits," Electrotries, August 1939, p. 14. See also 
A. J. Young: British Patent 435816; E. L. C. White: British Patent 455497; and 
H. M. Lewis: British Patent 530227. 



TRIGGER CIRCUITS 



61 



The discharge action introduced by the increase of suppressor 
potential and the resultant increase of anode current is obviously also 
a cumulative one, since the anode and screen currents are out of phase, 
and the initiation of the discharge is therefore rapid. 

If the suppressor grid takes a large forward current when it is driven 
positive, and if C 2 is sufficiently small by comparison with C lt then 
the suppressor grid will become negative before the completion of the 
discharge due to the condenser C 2 becoming charged by suppressor- 
grid current. This action has the effect of cutting off the anode current 
and increasing the screen current so that the screen finally takes all the 
current and the main condenser proceeds to recharge. On the other 



OKT.+ 



OQut 




—oH.T- 



Fig. 37. The Fleming-Williams time base 



hand, if C % is relatively large, the fall in anode current during discharge 
eventually produces a fall in screen potential, and the trigger action is 
again brought into play, though somewhat later. It is thus possible to 
obtain a saw-tooth wave-form across the condenser C v The rapidity 
of change-over from the charging to the discharging condition is 
determined by the amount of amplification existing between the 
suppressor and screen, together with the values of the stray capacitances. 

The circuit works best with a valve having a high ratio of anode to 
screen current so that, during discharge, the screen potential rises to 
a high value and accelerates the discharge since the higher the screen 
potential the greater is the anode current. With such a valve the 
cathode current increases considerably during the discharge, and 
square pulses will appear across a small resistance inserted in the 
cathode circuit. 

Figs. 38 and 39 show the wave-forms appearing between earth and 
the anode and screen of the valve. These wave-forms refer to the 
circuit of Fig. 37, for which representative circuit components are given 
°n the next page. 



62 





TIME 


BASES 


«1 


10,000 ohms 


H.T. 300 to 700 volts 


R* 


o*2 to 2-o megohms 


V Cossor MS/PEN 


*3 


0*05 to 0-5 megohm 




*4 


to 0-25 megohm 




c* 


0-1 J/.F. 




c 2 


i-o uT. 





TRIGGER CIRCUITS 



63 



By adjustment of i? 2 and i? 4 the charging time, and hence the time 
duration of the saw-tooth, is made controllable. The adjustment of 
R 3 controls the ratio of the charge to the discharge time, and hence 
determines the mark-space ratio for the square wave-form of screen 
voltage. The voltage on the screen grid when the charging and dis- 
charging times are equal is shown in Fig. 39. 




LrurL_TL 



Time 



Time — *- 
Fie. 38. Fig. 39. 

Voltage wave-forms pertaining to the circuit of Fig. 37 

Fig. 40 shows a method of explaining the behaviour of the time base, 
when using an EF50 valve, by means of a family of static characteristic 
curves taken with a potential of 350 volts fed to the screen through a 
load resistance of 50,000 ohms and with zero grid volts. These are not 
ordinary valve characteristics, since they are obtained by varying the 
potential applied to the suppressor grid and measuring its effect on the 
potential of the screen, which is connected through a resistance to a 
fixed H.T. supply, the condensers C x and C 2 being omitted. It will be 
seen that, where the curves are rising steeply, a small change in the 
suppressor potential produces a larger change in the screen potential. 

In the time base circuit the screen and suppressor are linked through 
the condenser C 2 and, so long as the charge on this condenser remains 
constant, there must be a constant difference between the screen and 
suppressor potentials. This condition is represented by drawing a 

straight line such as BGFC at a slope corresponding to g2 = i (i.e. 

dEgz 
at 45 if the scales are equal). Since the screen and suppressor poten- 
tials must satisfy the two equations represented by the static curve and 
the dynamic straight line, the stable potentials must be as indicated by 






the points of intersection. When the line cuts a curve at three points 
it would appear that there are three possible values for the screen 
potential, but the middle one is a point of unstable equilibrium and 
only the other two need be considered. 

Assume now that the condenser C x is charged to 180 volts. In this 
condition the suppressor and screen potentials are indicated by the 
point B on the curve, the screen potential being a minimum because 
the current in R z is a maximum. The anode then commences to take 
current and the screen current falls, thus causing the screen potential 
to rise to the point C: this occurs very rapidly due to the cumulative 




-100 



•o 



-50 



25 

Yo/ts 



SO 



75 



Fig. 40. 



-25 

Suppresso 

Characteristic curves for Fleming- Williams' time base 
(EF50 pentode with zero grid volts) 



too 



action. The locus then follows a curve such as CDE, reaching the 
point A when the condenser C x is discharged. The locus follows the 
dotted portion of the curve as the coupling condenser C 2 charges, 
while the suppressor is positive. During the period when C x is charg- 
ing, the screen potential remains constant while the suppressor potential 
slowly changes to the value B due to the condenser C 2 discharging 
through i? 4 and through the screen and cathode. 

This time base can be synchronized with either positive or negative 
signals. When it is desired to synchronize the time base from a negative 
signal the latter should be applied to the grid. While the condenser C x 
is being charged there is no anode current and the application of a 
negative pulse to the control grid reduces the screen current so that 
the screen potential rises. The suppressor potential is therefore caused 
to rise and anode current commences to flow, thus introducing the 
cumulative change of regime and discharging the condenser. When 
the synchronizing potential is positive it should be applied to the 



6+ 



TIME BASES 




suppressor grid. In this case, the suppressor potential is increased and 
the change of regime is introduced as before. In either case, the 
amplitude of the synchronizing potential should be kept to the lowest 
possible value consistent with effective control. This is especially 
necessary when the synchronizing potential is applied to the grid, 
since, if the signal is too large, it will cut off the screen current entirely. 
Furthermore, the synchronizing potential should be applied for a 
period which is only just sufficient to ensure effective control of the 
commencement of discharge. 

The Screen-coupled Transitron Flip-Flop 

A flip-flop,* as already stated, is a circuit having one stable and one 
unstable limiting condition. The movement from the stable state is 



-Q//J+ 



'•*3 



■- — VWVV— | — 1| — 



Control 
Potential <- Kl 



v 



OOut 



-QMI- 



Fig. 41. The screen-coupled transitron flip-flop 

effected by the application of a short-duration signal which releases 
a cumulative trigger action. This drives the circuit condition to the 
opposite extremity; i.e. to the unstable limiting condition. After a 
period, which is determined by the time constant of the circuit, the 
trigger action operates in the reverse direction and resets the circuit. 
The difference between a flip-flop and a continuously operating 
relaxation oscillator is that, in the case of the flip-flop, the fixed supply 
and bias potentials are so proportioned that oscillation is inhibited after 
each cycle, a new cycle being commenced by the application of a further 
initiating signal which momentarily brings the potentials to a point 
which, if maintained, would permit oscillation to continue indefinitely. 
Thus, a flip-flop is, in general, a relaxation oscillator which is over-biased 
(to or beyond cut-off). 

The expression "flip-flop" provides an excellent name for the 
arrangement since it describes the action perfectly. The circuit may 
be said to reside in the stable condition, to flip to the unstable regime, 

* A transitron flip-flop is also known as a "trant." 



TRIGGER CIRCUITS 



65 






and then to flop back to the stable regime after a predetermined lapse 
of time. The words flip and flop thus indicate the changes of regime 
and not the regimes themselves. 

The transitron relaxation oscillator shown in Fig. 37 may be converted 
into a flip-flop by applying a negative bias to the suppressor or by re- 
ducing the value of R 2 . The condenser C x should also be removed if it is 
desired to obtain a steep-sided output pulse at the anode. The principle 
of the circuit is essentially the same as the two-valve multivibrator type 
of flip-flop which is described later, except that the necessary amplifica- 
tion without phase reversal is obtained in a single valve, one of the 
couplings being the electronic one between the suppressor and screen, 
and the other the resistance-capacitance coupling provided by the 
condenser C 2 and the resistance # 4 , as shown in Fig. 41. Suitable 
values for the components are as follows: 



H.T. 250 to 350 volts 
i?j To suit the input circuit, 
say, 1,000 to 5,000 ohms 
R 2 10,000 ohms 
i? 3 50,000 ohms 



i? 4 200,000 ohms 

i? 5 normally zero (see page 

68) 
C 2 0-001 i*F. 
V EF50. 



The behaviour of the circuit can best be explained graphically by 
means of the static relationship between the suppressor and screen 
potentials. These relationships are plotted in Fig. 42 for an EF50 
valve operating under the conditions enumerated above, but with an 
H.T. supply potential of 340 volts. 

Several values of control grid bias are taken in order to determine 
the effect of the input pulse. The curves are obtained by applying a 
variable potential to the suppressor and measuring the resulting screen 
potential and suppressor current. A correction is made to the screen 
potential curves to allow for the effect of the suppressor current since, 
in practice, a part of this also passes through the screen resistance. 

The suppressor current is plotted in the lower family of curves, and 
it will be noted that the current is reversed above 40 volts. The effect 
of the current on the screen circuit is, therefore, to lower the screen 
potential in the region from o to 40 volts on the suppressor, and to 
raise it when the suppressor potential is above 40 volts. At still higher 
suppressor potentials, of the order of 100-120 volts, the suppressor 
current again becomes positive and rises rapidly with further increase 
of suppressor potential. 

The dynamic relation between screen and suppressor potential may 

be indicated by a straight line such as ABC drawn at the proper angle, 

i.e. 45 when the scales are equal, to correspond with the external 

coupling between the screen and suppressor, so that a given change 

7 



66 



TIME BASES 






in screen potential results in a similar change in that of the suppressor. 
In the circuit of Fig, 41 the presence of the condenser C 2 makes the 
change in suppressor potential equal to the change in screen potential, 
neglecting for the moment variations in the D.C. charge on the con- 
denser which can be represented by parallel movements of the line 
such as, for instance, to the dotted line DE (Fig. 42). 




•40 -20 *20 
Suppressor Volts 

Fig. 42. Characteristic curves for the screen-coupled transition flip-flop 

As the suppressor and control grid are both normally at cathode 
potential, the point A indicates the starting-point and the line ABC 
determines the operations as long as the condenser charge remains 
constant. While the static curve gives the screen potential which 
would result from the screen and suppressor currents alone, the 
operating line gives the actual screen potential which is a result of the 
sum of the currents flowing into the screen and into the stray capaci- 
tances. When the operating point lies above the static curve, part of 
the screen current is being supplied from the stray capacitance so that 
the potential must fall rapidly and vice versa. In this way it can be 
shown that the points of intersection at A and C are both stable, apart 
from variations in the charge on C 2 , while the intersection point at B 
is unstable. 

When a small positive pulse is applied to the control grid the static 
curve is lowered so that it no longer intersects the operating line in the 
AB region and the operating point moves very rapidly to a point such 
as C, returning to C at the end of the pulse. Reference to Fig. 43 (a), 






TRIGGER CIRCUITS 



67 



which shows the appropriate wave-forms, will make it clear that the 
suppressor is now negative, and the condenser C 2 discharges through R 3 
and i? 4 while the operating line moves to the position DE. At D the 
operating point must leave the static curve and another rapid movement 
occurs, ending at E. The suppressor is now positive and C 2 charges 
while the operating point follows the static curve from E to A and 
remains there since the voltage drop across i? 4 has then reached zero 
and A is a stable point. 

The movement from D to E is extremely rapid and is the period 
during which the anode current changes from zero to a large value. 
The period from E to A is one in which the condenser C 2 settles down 



Grid 
Potential 



Grid 

Potential 



—f 



J 



T 



Anode 
Potential 



Screen 



r~ 



Potefltial 



Suppressor 
Potential 



(b) 



Screen 
Potential 



Waveform with Diode 
Shunting Suppressor 




(C) 




Screen 

Poten tial 



Waveform with Negatm 
Bias on Suppressor 

Fig. 43. Voltage wave-forms for the screen-coupled transitron flip-flop 

to the stable condition, after which the circuit is ready to receive a 
further initiating pulse. If another pulse is applied before the point A 
is reached, the duration of the next unstable regime is somewhat 
reduced, depending upon the state of charge of C 2 at the moment of 
arrival of the pulse. 

It will be seen that the condition which makes the circuit give a 
single stroke instead of a continuous oscillation is that the suppressor 
is not biased on the unstable part of the static curve (i.e. the part which 
is steeper than the operating line). 

Referring to the wave-forms of Fig. 43 {a) for the positive input pulse, 
it will be seen that the flat-topped screen potential wave corresponds 
to the horizontal region CD on the static curve, which occurs during 
anode current cut-off. The hump at the end of the stroke is the depress- 
ion at E, which is a result of the addition of the suppressor current 
to that of the screen. The static curve, without the modification for 






68 



TIME BASES 






suppressor current, is shown (dashed) to the right of the point A. By 
introducing R 5 in series with the suppressor, the hump can be practi- 
cally eliminated owing to the reduction in suppressor current but 
further complications are introduced. 

When a negative impulse is applied to the control grid, the operating 
point moves to a point such as A' and returns to A at the end of the 
input pulse, unless it lasts sufficiently long for the condenser charge to 
alter appreciably. If the input pulse is small so that A' corresponds 
with a suppressor potential between zero and +40, the suppressor 
current is always in the same direction as the current through i? 4 and 
thus it helps to alter the condenser charge. The operating line therefore 
moves to the left and, at the end of the pulse, it may be clear of the 
static curve in the AB region so that the operating point flies to the 
left of C. After this, the sequence is the same as for positive initiation. 

An increase in the amplitude of the negative pulse applied to the 
grid drives the point A' farther to the right and the suppressor current 
may neutralize the current through i? 4 if the suppressor reaches the 
reversed current region. In this event the condenser cannot discharge 
and the circuit will not trigger, however long the pulse may last If 
R± is too high the reversed suppressor current may exceed the forward 
current through i? 4 so that the condenser discharges instead of 
charging. In an extreme case the condenser may be charged 40 or 
50 volts in the period of the input pulse and, at the end of the 
pulse, the suppressor is left 40-50 volts positive, thus allowing the 
suppressor current to maintain the charge so that the circuit becomes 
"locked." 

When i? 5 is introduced the locking trouble is greatly increased. 
The negative suppressor current is then able to maintain the suppressor 
positive and prevent the circuit functioning whenever the suppressor 
is momentarily driven into reversed current. 

Values of R B from 0-05 to 1 -o megohm are particularly apt to give 
this trouble with the EF50, but higher values such as 2-0 megohms 
give satisfactory results as the suppressor potential cannot then reach 
the locking- point. Although the hump is removed in this way, the 
return to the stable condition is appreciably slowed down and it seems 
preferable to use a valve which has no reverse suppressor current 
(e.g. MSPen.B) and which is therefore not liable to lock. If a diode 
is connected between the suppressor and cathode (diode anode to 
suppressor), the suppressor can be prevented from going positive and 
R 5 can be of the order of o- 1 megohm without any risk of locking. 
The end of the square wave is then followed by a tail, as shown in 
Fig- 43 (*)■ 
At this point it is interesting to note what happens if the points A 



TRIGGER CIRCUITS 



69 



and E are made to coincide, as they do with zero suppressor bias and 
approximately 1-5 volts negative bias on the control grid. The points 
C and D also coincide. With a positive pulse on the grid the circuit 
triggers, but the return to the stable condition occurs at or before 
the end of the pulse without any delay. With a negative input pulse 
the circuit gives a square-wave output whose duration depends on the 
amplitude and duration of the input pulse. This occurs because the 
condenser C 2 receives a charge while the suppressor is driven positive 
by the input and the amount of this charge determines the time period 
before the return stroke. The same factor is to some extent operative 
even with zero bias: the duration of the output square-wave is not 



OH.T.+ 



0H.T.+ 




-oH.T.- 



FiG. 44. Method of synchronizing a 

screen-coupled transitron flip-flop 

by means of negative synchronizing 

pulses 




og/as- 



Fig. 45. Method of synchronizing a 

screen-coupled transitron flip-flop 

by means of positive synchronizing 

pulses 






entirely independent of the input, being slightly greater for negative 
initiation than for positive. t 

If operation on both positive and negative pulses is not desired, it is 
possible to bias the suppressor to a point on the flat portion of the 
characteristic between C and D, so that the return stroke is not followed 
by a tail, but the hump at EAB is then transferred to the top of the 
wave as in Fig. 43 (c). Initiation is by negative input only and the dura- 
tion of the input pulse has little effect, since the circuit is triggered at 
the beginning of the pulse. 

Two further methods of applying the initiating pulse are shown in 
Figs. 44 and 45. In Fig. 44 the valve is normally conducting and the 
initiating pulse must be in the negative direction. This results m a 
positive pulse at the anode and a negative one at the screen grid. Fig. 45 
shows a circuit which operates with a positive input pulse and the 
output pulses are negative at the anode and positive at the screen grid. 

The flip-flop circuits already discussed are known as screen-coupled 
circuits, but there is a further type in which the external coupling 
takes place between the suppressor grid and the cathode, the former 
being at a fixed potential and the latter changing in potential owing to 
the presence of a cathode load resistance. 



7° 



TIME BASES 



u 200 1/ 



The Cathode -coupled Transitron Flip -Flop 

Whereas a screen-coupled transitron functions by providing amplifi- 
cation between the suppressor grid and the screen grid without phase 
reversal, the cathode-coupled transitron flip-flop provides amplification 
with phase-reversal between the suppressor grid and the anode and 

phase-reversal without amplification 
between the control grid and the 
suppressor grid. Of the phase-reversals 
mentioned above, that between the 
anode and the suppressor grid occurs 
only during the negative-slope portion 
of the characteristic. 

The basic circuit is shown in Fig. 46 
and the static characteristic curve as- 
sociated with the circuit is plotted in 
Fig. 47, from which it will be seen that 
for grid potentials above 27 volts positive with respect to the negative 
supply rail, the slope of the curve becomes negative. 






Input 
Volts 



■ 22. tea a> 



n 

12, teo w 



-0+251/ 



-o//.r- 



Fig. 46. Basic circuit of the cathode- 
coupled transitron flip-flop as used 
for making the measurements shown 
in Fig. 47 



f$8 


- 






























M 








































I 








194 






























: 










































/ 








"vil9d 










































1 

■^136 
.<3 


















Slope — 0-182 mA/ VQ j t 
Negative Resistanc 




























el 






























= 5500 ohms 


/ 






1j 

&182 






























c\ 








•§178 

3 






















































































































174 
















































































170 
























| 
















































166 


1 


l 












Fk 








i 






f 


6 




n 
i 


1 


H 


9tS 


2C 
ffti 


I 


2- 

Vol 


t 


I\ 


? 


3 


? 


3 


S 


40 



Fig. 47. Characteristic curve pertaining to the circuit of Fig. 46 

Fig. 46 shows the actual potentials and resistance values employed 
when making the measurements embodied in the curve of Fig. 47. The 
valve is a type VR116 and, for the purposes of measurement, the 
feed-back arrangements which are necessary to form a practical circuit 
are omitted. These feed-back arrangements are described later. 






TRIGGER CIRCUITS 



71 




Synch, 



O+200V 



-o+28V 



C-200V 
o3ias + 



Fig 



48. The cathode-coupled 
transitron flip-flop 



Examination of Fig. 46 will show that the suppressor potential is 
determined by the value of the cathode load resistance and the current 
flowing through it and this is, in turn, determined by the grid potential 
with respect to the negative rail. A change in grid potential therefore 
produces a change in the opposite sense in the suppressor grid potential. 

When the grid potential is increased to about 25 volts positive, the 
suppressor grid commences to reduce the anode current. WTien the 
grid potential is raised to 34 volts the suppressor grid has become 
sufficiently negative to cut off the anode current, and the cathode current 
passes entirely through the screen grid circuit and the valve then 
behaves as a triode. In this particular case, the value of the negative 
slope can be seen to be —0-182 
milliamperes per volt, which corre- 
sponds to a negative resistance of 
5,500 ohms. Obviously, therefore, if 
the anode is coupled to the grid, the 
circuit is capable of generating os- 
cillations. Furthermore, examination 
of the curve of Fig. 47 will show that 
if the grid is biased to the points A 
or B on the curve, the circuit will 
have one stable limiting condition 

and will therefore function as a flip-flop; but, if it is biased to the 
point C, it will have two unstable limiting conditions and will operate 
as a square-wave generator (see Appendix VI). 

Fig. 48 shows the flip-flop form of the circuit, the condenser C 1 
being inserted to provide the necessary feed-back from the anode to 
the grid of the valve. Suggested component values are: 

R 1 20,000 ohms 

R 2 100,000 ohms 

R 3 10,000 ohms 

C x dependent upon requirements 

P\ VR116, 

The condenser C t provides positive feed-back from the anode to the 
grid during the negative slope portion of the characteristic and negative 
feed-back in the positive portions. Since there is phase reversal between 
the grid and the suppressor grid, the feed-back circuit causes the 
suppressor grid to become more negative when the anode goes positive 
during the negative slope portion of the characteristic curve. 

It is now difficult to obtain valves of the VR116 type, but their 
special merit for use in this circuit lies in the fact that the suppressor 
grid is closely wound so that the potential required at the suppressor 






7 2 



TIME BASES 






grid to cut off the anode current is small— some 5 to 10 volts dependent 
upon the potentials of the anode and screen grid. 

The value of the bias potential with respect to the negative rail can 
be determined from an inspection of Fig. 47, viz. +24 volts approxi- 
mately for a positive synchronizing signal or +36 volts approximately 
when the synchronizing signal is negative. 

In order to understand the operation of the circuit, assume that the 
valve is biased at +24 volts. The anode potential will then be approxi- 
mately 168 volts and the injection of a positive synchronizing potential 
will carry the grid potential beyond +27 volts, resulting in an increase 
in the anode potential which will rise to 200 volts due to the feed-back 
circuit causing the grid potential to rise with that of the anode. When 
the anode current becomes cut off at 200 volts, the time constant 
(R 1 ~{-R 2 ) C 1 commences to discharge and the grid potential therefore 
falls. When this potential falls below +34 volts, the anode potential 
commences to fall, thus reducing the grid potential still further until 
finally the grid potential reaches the point A once more and the circuit 
becomes stable. 

This circuit is due to B. C. Fleming- Williams, of A. C. Cossor, Ltd. 

TWO-VALVE TRIGGER CIRCUITS 

The Two-Valve Flip-Flop 

There are a number of variations of the two- valve flip-flop, and one 
form of the circuit is shown in Fig, 49. This circuit may be regarded 
as an over-biased multivibrator in which one of the condensers has 
been replaced by a direct connection. However, as will be explained 
later, the essential feature which makes the circuit a flip-flop is not the 
presence of the direct coupling, but the fact that sufficient bias is 
applied to prevent continuous oscillation. 

The application of a negative potential to the grid of V x results in 
the grid of V 2 being driven positive and thus causes anode current to 
flow in V 2 which is normally biased to, or beyond, cut-off. The screen 
potential of F x is therefore depressed so that the grid of V z is held 
positive during and after the cessation of a short- duration exciting 
signal. The circuit is returned to the normal condition by the dis- 
charge of the coupling condenser. The duration of the unstable regime 
may be controlled by adjustment of the coupling time constant. The 
circuit, in essence, provides A.C. feed-back in a D.C. amplifier in the 
sense that the potential appearing at the grid of V 2 appears also at the 
anode of F 2 after amplification by the valves V 2 and V v This is followed 
by A.C. feed-back from the anode of V 1 to the grid of V 2 . Since the 
complete amplifying chain is continuous, it is immaterial whether the 






TRIGGER CIRCUITS 



73 



exciting signal is applied to the grid of V x or of V 2 provided that it is 
of the appropriate polarity and has sufficient amplitude. The output 
signal may be taken from the anode of either valve. 

Appropriate values for the components of Fig. 49 are as follows : 

i? 2 50,000 ohms to 1 megohm (dependent upon the required 

time constant) 
R*, R* 50,000 ohms. 



L 3> 



The value of R 5 ^-R 6 must be so arranged that the current J through 
the two resistances is sufficient to make IR & greater than the potential 




■OH1+ 



Control 
Potential 



H.T- 



Fig. 49. Two-valve flip-flop with one A.C. coupling 



required to cut off anode current in V 2 . The time constant C z R b must 
be at least ten times the value of the time constant C X R 2 . 

It should be noted that an essential difference between the two-valve 
flip-flop and the multivibrator is that the flip-flop has one of the valves 
biased to cut-off. This bias is capable of keeping the circuit in its 
stable condition after one period of operation, even if two A.C. couplings 
are used between the valves. 

An example of a flip-flop using two A.C. couplings is shown in 
Fig. 50. The interval between the flip and the flop, i.e. the duration 
of the unstable regime, depends upon the time constants of the two 
coupling circuits in very much the same way as in the multivibrator. 

An important point to note is that if the synchronizing, or exciting 
pulse is applied to the grid of V x it must be in the negative direction 
with either form of two-valve flip-flop and must last for a shorter time 
than the unstable regime. If it is required to operate a flip-flop from 
a positive pulse it should be applied to the grid of F 2 , although, in this 
case, a larger pulse amplitude is generally required. It should be noted 
that this arrangement is not the same as that of the transitron flip-flop 
which will operate, though in a somewhat different manner, with either 
a positive or a negative pulse applied to the same electrode. 






74 



TIME BASES 



Many interesting variations of these circuits are in use and similar 
results can, of course, be obtained when using pentodes, by coupling 
the suppressor grids to the anodes. Alternatively, the initiating pulse 
may be applied to the suppressor instead of to the grid, as shown in 
Fig. 51, in which the screens are separately supplied. They may, in 



OH.T.+ 




Control 
Potential 



■OKI- 



Fig. 50. Two-valve flip-flop with two A.C. couplings 

practice, be joined together and fed from a single potentiometer con- 
sisting of two fixed resistances with a condenser connected between 
the screens and the negative supply rail. 

The wave-forms appearing at various parts of the circuit of Fig. 51 
are shown in Fig. 52. Since a negative initiating signal is applied to 



GKT.+ 



Control 
Potential 




OHJ- 



Fig. 51 . Two-valve flip-flop in which the initiating pulse is applied to the 

suppressor grid 

the suppressor grid of the valve V lt the anode potential of this valve 
rises rapidly and the potential of the grid of F 2 is also driven in a 
positive direction up to the point at which it becomes limited by grid 
current. As a result of this, the anode potential of the valve V 2 falls 
and drives the grid of V x negative. This sequence of operations "com- 
pletes the "flip." The negative potential thus applied to the grid of 
V-y continues to cut off the flow of anode current until the condenser 
C 2 is discharged sufficiently to permit a small anode current to flow. 






TRIGGER CIRCUITS 



75 



— if f 



Control Pulse on 
Suppressor of V t 



Anode of Vi 



J V-V 



Grid of V 2 



Anode of V 2 






This occurs at the end of the exponential portion of the curve at 
the bottom of Fig. 52. The procedure is then reversed, the potential 
of the anode of V x returning to its normal level. This procedure, which 
constitutes the "flop," is indicated by the following vertical portions of 
the same figure. 

An interesting point to be noted in connection with the flip-flop is 
that the circuit may be forcibly returned to the stable condition 
before the end of the normal 
duration of the unstable regime. In 
order to carry this into effect the 
circuit of Fig. 5 1 may be employed, 
in which the initiating negative 
pulse is applied to the suppressor 
of V t and the arresting negative 
signal should be applied to the 
suppressor of V %y which should, in 
this case, be connected to the 
cathode via a leak resistance. When 
the circuit is in the unstable con- 
dition, the valve V % is conducting 
and, therefore, the application of a 
negative pulse will cut off the 
anode current and drive the grid of 
V x positive so that the "cumulative 
action is engendered in the reverse 
direction, thus returning the circuit 
to the stable condition. 

Most flip-flops have the charac- 
teristic that they will not flip 
satisfactorily immediately after they 

have flopped. The reason for this is that, although the circuit is in a 
stable condition, the condensers require a certain time to return to 
their normal state of charge. This applies particularly to the condenser 
C x since, at the flop, it is driven negative, usually far beyond the value 
of cut-off potential for the valve, and, unless time is allowed for the 
charge to leak away, the initiation of a new flip will depend upon the 
instantaneous value of the remaining charge. 

Allen* has developed a method of overcoming this difficulty by the 
addition of an inverted diode connected between the grid of V 2 and 
earth as shown in Fig. 53. The presence of this diode prevents the possi- 
bility of the grid of V 2 being driven more negative than the potential 
applied as a fixed bias to the valve. The circuit is thus made ready to 

• A. Allen: Research Department, A. C. Cossor, Ltd., London. 



Grid of V x 



Fig. 52. Flip-flop wave-forms 



7 6 



TIME BASES 



accept another initiating signal almost immediately upon the cessation 
of the unstable regime. 

The circuits of Figs. 49, 50, 51 and 53 can each be modified by 
taking the grid leak resistance R 2 to the cathode of V % instead of to the 
earth rail. As a result of this, the potential across the condenser C x 
does not require to discharge to such a large extent before current is 
again initiated in the valve V % . The time during which anode current 
flows in V 2 is therefore reduced and the precision of timing is somewhat 
increased since anode current commences to flow while the condenser 
is discharging at a more rapid rate. A still more rapid rate of discharge 



*i 



C 2 



Control 
Potent/a/ 



la t 



T 



-OH.T-* 



T 

I 






4 



c 3 



•* » * 0//.Z- 

FlG. 53. Allen's flip-flop circuit allowing for early reinitiation 

and, hence, greater precision of timing and freedom from jitter (see 
page 34) is introduced if the grid leak resistances R r and R z are taken 
to the positive H.T. supply rail. 

One can now imagine a further case in which a circuit is able to 
remain in either of two stable limiting conditions for an indefinite 
period so that it may be switched from one to the other condition at 
will. One of the best-known circuits of this type has been developed 
by Eccles and Jordan. 

The Eccles -Jordan Trigger Circuit 

The Eccles-Jordan circuit,* which was actually developed prior to 
the flip-flop, is similar to the multivibrator, but there are D.C. couplings 
between the anodes and grids, and, as a result, it has two stable limiting 
conditions instead of two unstable limiting conditions as in the multi- 
vibrator. Since the circuit is essentially stable it requires an external 
exciting signal to cause it to change from one stable condition to the 
other. The circuit is operated by a negative signal which is usually 
applied to the grids of both valves, as shown in Fig. 54. In this way 
the valve which happens to be conducting at the moment is cut off by 

* W. H. Eccles and F. W. Jordan: Radio Review, 191 9, 1, p. 143. 









TRIGGER CIRCUITS 



77 



the cumulative action of the currents in the two back-coupled valves 
upon the application of the negative pulse. The circuit in the figure is 
an improvement due to Reich,* in which pentodes are used instead of 
triodes, thus permitting the suppressor grids to be used for the trigger- 
ing action and leaving the grids free for the application of the initiating 

signals. 

The resistances R t and R t are inserted to provide suitable values of 
potential at the grids of the valves by potentiometer action, the source 
of potential being, in effect, that of the anode of the opposite valve. 



H.T.+ 




Control % p 
Potential 



O/f.K- 



Fig. 54. Eccles-Jordan trigger circuit (improvement due to Reich) 

The condensers C x and C 2 are added to enable the steep wave-front 
of the signal at the anode of each valve, at the instant of change-over, 
to appear at the grid of the other valve without appreciable distortion. 
These condensers therefore counteract the effect of stray capacitance 
and allow the circuit to be operated by an exciting pulse of very short 
duration: their value, which is very small, is best found by trial and 

error. 

In view of the fact that both grids are driven negative to the same 
extent by the control pulse, it is not immediately obvious why the 
two valves change their role at each exciting pulse. To find the explana- 
tion of this it is best to consider first what conditions are necessary to 
permit a cumulative trigger action to take place and finally to examine 
the way in which the control pulse brings about these conditions. 

It is clearly impossible for a trigger action to occur in a circuit of 
this kind so long as one or both valves are completely cut off. Imagine 

*H J Reich- "A New Vacuum Tube Counting Circuit," Review of Scientific 
Instruments. 1918. 9, p. 222; V. H. Regener: "Design and Use of Directly Coupled 
Pentode Trigger Pairs," ibid., 1946, 17, p. 180. 



78 



TIME BASES 



that one of the inter-valve couplings is disconnected at the suppressor 
grid and that the latter is temporarily connected to a source of variable 
potential. If, now, a small disturbance of this potential produces a 
larger disturbance at the point which is normally connected to the 
suppressor grid, the conditions are correct for producing a cumulative 
action since the overall gain of the two valves, considered as an amplifier, 
must exceed unity. This condition is easily attained even when the 
valves are biased near to cut-off. 

The condition for obtaining a trigger action in the Eccles-Jordan 
circuit may now be visualized as follows: it is necessary that some 
anode current should flow in both valves. 

Now consider the effect of the control pulse : when a negative signal 
is applied to both grids it has no direct effect upon the valve, say V l7 
which is already cut off by the negative suppressor potential. On the 
other hand, the negative signal reduces the anode current of the valve 
V 2 and the resulting rise of anode potential brings the suppressor of 
the valve V x to a point where anode current just begins to flow. Since 
anode current is now flowing in both valves, a cumulative process 
takes place and the valve V % is cut off completely while the anode 
current in F l5 which was originally cut off, cannot be immediately and 
completely restored to the cut-on condition because the grids are still 
negative. In fact, unless the trigger action occurs precisely at the peak 
of the control pulse, the anode current of V x must first rise suddenly 
and then fall at a rate which depends mainly on the shape of the control 
pulse and then it will finally rise again as the control pulse tails away. 
This is illustrated in Fig. 55, which shows the anode potential wave- 
forms for a complete cycle of two operations. At the first exciting 
pulse, the dotted line A indicates the simultaneous rapid changes which 
represent the trigger action. At the second pulse, the dotted line B 
indicates that the peak of the pulse is coincident with the starting-point 
of the second downward movement of the anode potential of the valve 
which is to conduct during the next half-cycle, in this case V 2 . The 
downward overshoot at C is brought about by the fact that the 
suppressor becomes less positive as the charge on the coupling con- 
denser, in this case C 1( reaches equilibrium. As pointed out before, the 
condensers C x and C 2 are very small in value and the time constants 
C X R & and C 2 7? 6 should be of the order of the duration of the exciting 
pulse. 

From the above remarks, it is evident that the rise of potential on 
the anode of V 2) due to the switching process at the time A, must 
occur before the control pulse reaches its crest, whereas the fall of 
potential on the anode of V x will take place in two stages, the end- 
point being synchronized with the end of the control pulse. If the 






TRIGGER CIRCUITS 



79 



control pulse amplitude is larger than necessary, e.g. if it is large enough 
completely to cut off both valves for a short time, no difference will 
be observed at the instant A, but the zigzag effect will be more pro- 
nounced, as shown by the dotted curve. Provided the duration of 
the control pulse is short compared with the time constants of the 
inter- valve couplings there is no tendency for a second trigger action 
to occur at the end of the pulse. 



Zero — 



V 



\y 



Grid Potential 
Vj_ and V 2 



Cut-off' 



1 



L 



V x Anode Potent/a/ 



Cut-cff- 



u 



V 9 Anode Potential 



Fig. 55. 



I C 
A B 

Wave-form diagrams for the Eccles-Jordan circuit 



Certain valves, in particular the EF50 type of valve, exhibit a 
negative suppressor current when the suppressor potential is driven 
more than about 40 volts positive with respect to the cathode. This 
feature is likely to prevent satisfactory operation of the circuit of 
Fig. 54, but the difficulty may easily be avoided by applying the negative 
exciting pulses to the suppressors and by using the grids for the 
cross-couplings. 

The condenser C 3 has the effect of holding the value of the potential 
across i? 7 constant. This potential is used to provide cathode bias for 
the two valves by means of the current flowing through the "bleeder" 
resistance R B . It should be noted that this bias potential is applied only 
to the suppressor grids and not to the control grids. 

A similar circuit has been used by Starr* for making peak voltage 
measurements. 

In essence, this circuit forms a convenient means of comparing the 

* A. T. Starr: "A Trigger Peak Voltmeter using Hard Valves," Wireless Engineer, 
10 35. 12. p-601. 



8o 



TIME BASES 



peak potential of a wave-form with a measured steady potential which 
is obtained from an adjustable potentiometer connected across a battery. 
The potentials to be compared are connected in series with the grid 
and grid leak resistance of one of the valves and measurements are 
made of the steady potentials required to cause the circuit to trigger 
with, and without, the wave-form under test. The difference between 
the two values of the steady potential is the peak potential of the 
wave-form. 

Evans' Trigger Circuit 

Evans has developed an interesting trigger circuit* of the type which 
has two stable limiting conditions. The circuit is shown in Fig. 56 in 
which R 2 is a resistance across which a triggering potential is developed. 








HT- 



'VWW -1 

Rs A 



Fig. 56. Evans' trigger circuit 



The circuit was originally devised for use at radio telegraph receiving 
stations for squaring signals (which arrived in a distorted form) before 
applying them to a tape recorder. 

The value of R 2 is so adjusted that the current flowing through it is 
just sufficient to cut off anode current in V x . In these circumstances, 
there is no current in R A , and hence the anode current in V 2 is at a 
maximum value. Upon the application of a positive steady potential 
to the grid of V 1 anode current commences to flow in V x and this 
reduces the anode current in V z by the application of bias to its grid. 
The anode current of F 2 , flowing through R 2 , increases the positive 
bias on the grid of V x . Thus, if the loop gain over the complete circuit 
is greater than unity, the effect is a cumulative one and the circuit 
triggers extremely rapidly to its second stable condition of maximum 
anode current in Vj and zero anode current in V t . Upon the reduction 
of the input signal to a lower value, which depends upon the circuit con- 
stants, the triggering action reverses and the circuit is restored to its 

# H. Faulkner and G. T. Evans: "Some Developments in Telegraph Technique 
as applied to Radio Circuits," Journ. P.O.E.E., Reprint No. 136, 1931; G. T. Evans 
and L. T. Arman: "A New High-speed Multi-Channel Carrier Telegraph System," 
Journ. P.O.E.E., 1935, 27, p. 241. 



i 



TRIGGER CIRCUITS 



81 



Output 
Potent/a/ 




initial condition. The current in R 6 thus falls to zero upon the applica- 
tion of a positive input signal of a value depending upon the circuit con- 
stants and rises again to a maximum when the input signal has fallen to 
another predetermined value. Normally, when the input potential is 
increasing, a larger potential is required to produce the triggering action 
than when it is decreasing. There is thus a "backlash" effect (see 
Fig. 57, solid curves), but by suitable adjustments to the value of R z 
the two triggering potentials may be made almost to coincide. 

For some purposes, e.g. repeated telegraph signals, it is occasionally 
required to arrange that the reverse triggering action should occur at 
a higher potential than the original triggering action, as shown in the 
dotted curve of Fig. 57. This effect may be obtained by the presence 
of the network R 1 C 1 which is omitted when the effect is not required. 
When R X C X is present, and the first trigger action has taken place, grid 
current flows in V x and the con- 
denser becomes charged and adds 
a component in the opposite sense 
to that of the incoming positive 
signal. Note that this component is 
not added until after the triggering 
action has taken place, and that it 
can only occur when the input 
signal is larger than is necessary to 
initiate the triggering action. As a 

result, the grid potential does not require to fall to so large an extent to 
initiate the reverse triggering action as is the case when R 1 C 1 is 
omitted. 

Schmitt 's Trigger Circuit 

Schmitt* has developed an interesting trigger circuit which is a more 
modem version of that due to Evans. The circuit arrangement is 
shown in Fig. 58, from which it will be seen that there are two back- 
coupled triodes having direct couplings. One of these is a resistance 
coupling between the anode of the first valve and the grid of the second, 
while the other is the common cathode resistance i? 5 . The circuit 
operates with a change of steady input potential. 

The circuit triggers in one direction when the input potential is 
raised to a critical value and triggers in the reverse direction when the 
input is reduced to another level. The two levels may be made to 
approach one another by reducing the value of R 5 . If i? 5 is made 
slightly smaller than that required to make the two levels coincide, the 
device is no longer a trigger circuit but an extremely sensitive amplifier. 
* O. H. Schmitt: "A Thermionic Trigger," Journ. Sci. Insts., 1938, 15, p. 24. 
8 



Fig. 57. 



'Input Potent ial 



Characteristic curves pertaining 
to the circuit of Fig. 56 



82 



TIME BASES 



The amount of back-lash, i.e. the difference between the two operating 
potentials, may be smoothly adjusted to lie between 01 volt and 20 volts. 
The input current is less than one microampere and the output current 
may reach 20 milliamperes at 200 volts. The sensitivity of the circuit 
is very high and it may be used to trigger a time base, but it has many 
other applications; for instance, it may be used to operate a relay placed 
in the anode circuit of the second valve. 

O//J+ 



R* 



... 



it r» 

f 1 Li 



Control 
Potential 



r\ 



R3 




■oti.T- 



-&Bias-200V. 



a, 

R 3 
R A 
R. 



Fio. 58. Schmitt's trigger circuit 

The values employed by Schmitt are given below: 

o*i megohm 

0-5 megohm 

2-0 megohms 

10,000 ohms 

50-1,200 ohms 
V x and F 2 6A6 (any ordinary triodes may be employed) 
H.T. 200 volts 
Bias — 200 volts. 

When the input potential applied to the grid of the first valve is 
zero, maximum current flows in the second valve and the anode current 
of the first valve is cut off by the potential drop across R 5 . If, now, the 
potential on the grid of V 1 is increased by an amount which is sufficient 
to overcome the potential drop across R h and to permit some anode 
current to flow in V ± , the grid potential of V z will become sufficiently 
negative to reduce the anode current of V 2 considerably, thus reducing 
the value of the potential E c (across R 5 ) which, in turn, will increase 
the grid potential of V v The effect is thus a cumulative one and a 



TRIGGER CIRCUITS 



83 




rapid change from one regime to the other is obtained. A small capaci- 
tance C 1 may be connected across R 2 to increase the velocity of the 
trigger action and to allow the circuit to function with an input signal 
consisting of a brief pulse. 

The output may obviously be taken from either anode or, if desired, 
from both anodes, in which case push-pull operation of any device 
connected thereto is available. 

This circuit is an interesting example of the use of positive feed-back 
to reduce the effective grid base of the valve V x and even to make it 
negative. Further information on the operation of the circuit is given 
by Williams,* who suggests that R h should be made sufficiently high 
in value to ensure that the common cathode potential resides at about 
one-half of the total H.T. supply potential, the latter being about 300 
volts. The adjustment for the most rapid switch-over from one stable 
condition to the other is then obtained by making the value of R x 

rather greater than is necessary to satisfy the equation R 1 = — , where g 



is the mean slope of the two valves in amperes per volt. It should be 
noted that this provides the most rapid switch-over, but does not adjust 
the potential values at which the switch-over takes place. The effective 
switch-over difference of potential (lying in the neighbourhood of half 

. 1 /£, *\ v H.T. potential 
the H.T. potential) is approximately— I -_— -- 1 X — 2 ~~' 

i.e. this represents the difference between the two switching potentials. 

Flip-flop Form of Schmitt's Trigger Circuit 

It will be obvious that Schmitt's circuit may also be modified by 
the employment of an R.C. coupling between the anode of V 1 and the 
grid of F 2 , in which case it will behave as a flip-flop. The resistance 
R 2 of Fig. 58 is replaced in Fig. 59 by the condenser C v and i? a is 
taken to the cathode of V 8 . 

With this circuit V 2 will normally conduct, since the grid and cathode 
will be at the same potential and the anode current will flow through 
R f and cut off anode current in V v Upon the application of a suffi- 
ciently large positive pulse to the grid of V lt anode current will com- 
mence to flow in that valve and the anode current of V % will be reduced. 
The potential across R & will fall, although this fall will be partly com- 
pensated by the rise of anode current in V v To obtain the most 
satisfactory results, the anode current at zero grid volts should be 
greater for V 2 than for V v 

This circuit may be modified by taking R 3 to the H.T. supply rail, 

* F. C. Williams: "Introduction to Circuit Techniques for Radiolocation," Journ. 
I.E.E., 1946, 93, Part IIIA. No. 1, p. 289. 



84 TIME BASES 

in which case it may be necessary to raise the value of the steady 
potential appearing at the grid of V v This may be carried out by 
connecting the grid of V x to the junction of two resistances of suitable 
value connected in series across the positive and negative supply rails. 
The control potential should then be applied to the grid of V 1 via a 
condenser. The steady potential between the negative supply rail and 
the grid of V lt Eg vl should be so adjusted that Eg vl — Eg V2 is greater 
than is required to cut off the anode current in V v 



o/f.I+ 




Control 
Potential « 



oNI- 



FiG. 59. Flip-flop form of the circuit of Fig. 58 

A circuit which is closely related to that of Schmitt has been used 
by Colebrook* as a sensitive D.C. galvanometer amplifier for bridge 
measurements. Colebrook replaced the resistance R 2 , in Fig. 58, by a 
battery and omitted R s . Then, when R 5 is reduced below the value at 
which the circuit fails to trigger, i.e. to about 15 ohms, the arrangement 
forms a very sensitive D.C. amplifier, and a galvanometer inserted in the 
anode of V % may be employed to indicate when a bridge is balanced 
or for a number of other purposes. 

Chance's Trigger Circuit 

Results which are similar to those obtained with Evans's and Schmitt's 
circuits are obtainable with a gas-filled diode. f The appropriate circuit 

is shown in Fig. 60. The operation 
of the circuit depends on the fact that 
the striking and extinction potentials 
have different values, as explained on 
page 15. A disadvantage of the cir- 
cuit is the necessity for a high value 
of input potential but, since the slope of the current characteristic 

*F. M. Colebrook: "A Valve Voltmeter with Retroactive Direct-Voltage Amplifi- 
cation," Wireless Engineer, 1038, 15, p. 138. 

f B. Chance: "Some Precision Circuit Techniques used in Wave-form Generation 
and Time Measurement, 1 ' Rev. Set. Imts., 1946, 17, p. 396. 



O WVW- 

Input 



m 



Output 



Fig. 60. Chance's trigger circuit 






TRIGGER CIRCUITS 



85 






of the diode is high, the output impedance is much lower than 
is the case with an ordinary vacuum diode. The series resistance 
must have a value sufficient to limit the anode current to a safe value. 

SECONDARY EMISSION TRIGGER CIRCUIT 

Black's Time Base 

Under certain conditions the anode current in a tetrode or screen-grid 
valve may be reversed in polarity. This effect is not usually found in 
modern valves which are generally designed to eliminate it, but it was 
quite common with the screen-grid valves which were on the market 
before the year 1930. A reversal of anode current means that more 
electrons leave the anode than arrive there from the cathode, bomeol 
the electrons which reach the anode after having passed through the 
spaces between the screen wires have sufficient velocity to dislodge 
electrons from the atoms of the anode material. If the screen is more 
positive than the anode, these secondary electrons are drawn back to the 
screen so that the net anode current is the difference between the forward 
current and the secondary emission current. If, on the average, each 
primary electron ejects more than one secondary from the anode, then 
the anode current will be reversed. This condition can only arise when 
the anode (emitting surface) potential is less than that of the screen and 
when this potential difference lies between two limiting values. *or 
example, given the conditions shown in Fig. 63, when the screen 
potential is 390 volts, the anode current is negative for values of the 
anode potential between approximately 65 and 340 volts, the potential 
difference then lying between 325 and 50 volts. This potential difference 
is reduced when the screen potential is lowered, and it is also dependent 
upon the material of which the anode is made and the condition of its 
surface. These conditions are those of the Dynatron described by Hull 
Under these conditions, the practical result is that the flow of anode 
current through an anode load resistance produces an increase of anode 
potential. The result of making the grid more positive is, in these 
circumstances, to increase the primary current from the cathode and 
therefore to increase the secondary emission current from the anode, 
with the result that the anode becomes more positive. 

Blackf has emploved this type of valve arrangement in a relaxation 
oscillator which gives a saw-tooth output. The fly-back is stated to 
be very rapid. In one form of the circuit, the condenser C x to be 
discharged is connected between cathode and screen of a suitable tetrode, 
and the anode is coupled to the control grid through a further condenser 

*A. W. Hull: "The Dynatron. A Vacuum Tube possessing Negative Electric 

RC | I^blad? A new Ho'rd Valve Relaxation Oscillator," Electrical Communication, 
July 1930. 



86 



TIME BASES 



C 2 , as shown in Fig. 61, for which the following potentials and com- 
ponents may be employed: 



Ex 


450 volts 


R 3 


2 megohms 


E t 


200 VoltS 


R* 


o-i megohm 


*. 


—30 volts 


Cx 


0*0005 to °"°S M'F 


*i 


1 megohm 


c\ 


0-0001 tO O-I [xF. 


** 


1 megohm 







When the circuit conditions result in a reversed anode current the 
rise of anode potential carries the grid positive due to the presence of 
C 2 . This, in turn, increases the current through the valve and therefore 
increases the anode potential still more. 



Outo 




Fig. 61. Black's time base 

Thus, a cumulative trigger action takes place and the grid is driven 
positive with great rapidity. The screen current is, of course, very 
high when the control grid is positive and the time base condenser C x is 
therefore rapidly discharged by the screen and anode currents. The 
fly-back produced by this time base is exceptionally rapid, and it is 
sometimes necessary to take steps to retard it; this may be done by 
inserting a small resistance in series between the condenser and the 
screen. 

The basic principle involved in the operation of this time base is 
easily demonstrated by means of the arrangement shown in Fig. 62, in 
which the coupling between the anode and grid takes the form of a 
battery which is to be regarded as equivalent to a charged condenser 
of very large capacitance. The potential of this battery should be be- 
tween 150 and 250 volts. The screen of the valve, which may be an 
Osram MS4 or Mazda AC/SG, is connected through a milliammeter and 
a switch to the H.T. supply which should be of about 400 to 500 volts. 
When the screen circuit is momentarily closed, it will be found that 









TRIGGER CIRCUITS 



87 



the screen current reaches a high value in the region of 100 milli- 
amperes, while the anode current is about 10 milliamperes. The switch 
must be only momentarily closed, otherwise the valve will be per- 
manently damaged. It is to be noted that, since the grid leak is 



C 200m/i 



=(y-~X 



o#.r+ 



X 



0-20tn/! 



Zht 



' 2 meg oh r. 



■Otf.T.- 



Fig. 62. Circuit for demonstrating the basic principles of Black's time base 




ZOO 300 

Anode Potential (Ea) 



400 
Volts 



Fig. 63. Static anode potential -anode current curves for an MS4 valve 

comparatively high in value, practically all the anode current passes into 
the grid and the grid potential becomes from 10 to 20 volts positive 
with respect to the cathode. The anode current does not tend to 
discharge the battery since the former is reversed in polarity. 

Some static characteristic curves of anode current plotted against 
anode potential for an MS4 valve are shown in Fig. 63. It will be seen 



88 



TIME BASES 



that the anode current is only reversed when the anode and screen 
potentials are within certain limits. Thus, if the anode potential is 
240, the anode current is positive for screen potentials below 280, and 
negative for screen potentials above that value. 

The operation of the circuit is as follows : the time base condenser C x 
is charged from the supply voltage E x through the charging resistance 
R v At a certain potential drop across C v the valve suddenly takes a 
heavy screen current which discharges C 1 until the screen potential 
reaches another critical point when the discharge suddenly ceases. 
While C x is being charged the anode and screen currents are both 
zero because the control grid is biased beyond cut-off. The bias control 
determines the screen potential at which screen current begins to flow. 
The screen potential at which the discharge ceases is usually about 
40 to 60 volts above the value of E 2 so that the output amplitude is 
increased by reducing the potential E 2 . However, the circuit becomes 
inoperative if E 2 is reduced much below 200 volts. The reason for this 
is evident from the characteristic curves (Fig. 63) since the reversed 
anode current becomes progressively smaller as the screen potential 
is reduced. 




o//.r+ 



CHAPTER AT 

BLOCKING OSCILLATORS AND INDUCTIVE 
TIME BASES 

There are various other forms of time base which have not yet been 
discussed. Among these are the following : 

(a) Blocking oscillators. (Transformer-coupled time bases.) 

(b) Inductive time bases. 

BLOCKING OSCILLATORS 

Appleton, Herd and Watson Watt's Time Base 

The first hard valve time base was of the transformer-coupled type 
and was developed about 1923 by Appleton, Herd and Watson Watt.* 
The circuit was known as a "squegging oscillator" or "squegger."f 

This circuit consists of a transformer-coupled valve F x , oscillating 
fairly violently at a radio frequency and having a condenser C x in the 
cathode-grid circuit. At each positive peak of potential at the grid, 
current flows from the transformer secondary winding through the 
valve V x from grid to cathode and 
into the condenser C 1( which thus 
accumulates a charge such that 
the mean grid potential becomes 
increasingly negative. When the 
negative potential reaches the cut- 
off bias potential of the valve, the 
anode current and, hence, also the 
alternating potential at the grid, is 
cut off and remains so until the 
charge on the condenser leaks away 

sufficiently through the diode to permit the resumption of oscillation. 
While the anode current is cut off, the condenser C x loses its charge 
via F 2 and the grid makes a potential excursion towards zero volts. 
Upon this excursion there is superimposed a damped oscillatory 
motion which is the second half-cycle of the oscillation. Originally, 
the diode was connected directly across the condenser C 1 with the 

• E V Appleton J. F. Herd and K. A. Watson Watt: British Patent 235254. 

+ The term ''squegger" was originally coined in the Royal Air Force where the 
circuit was used as a rough test for the values of grid leak resistances I .be squeak 
obtained with a standard resistance of known value was compared with that o the 
resistance under test. The higher the squeak frequency the lower the value or the 
resistance. Since, by squeaking, the circuit behaved m a "Megger, the circuit was 
known as a "Squegger." 

89 



H.T.- 



Out 



Fig. 64. Appleton, Herd and 
Watson Watt's time base 



90 



TIME BASES 









anode of the diode joined to the cathode of V v However, the in- 
ventors also described an improvement, shown in Fig. 64, in which 
the anode of the diode is connected to the positive supply rail. In 
this case, the ratio of the maximum to the minimum potential across 
the diode is much reduced with consequent improvement in linearity 
of the discharge of C x and hence of the trace. 

The charging time must be very much less than the discharging time 
if the fly-back time is to be short in comparison with that of the trace. 
This time ratio can be improved by reducing the temperature of the 
diode cathode in order to increase the impedance and also by ensuring 
that the potential swing applied to the grid of V lt when the valve is in 
the oscillating condition, is as high as possible, so that the condenser 
will be rapidly charged. To this end, it is advantageous to employ a 
tightly coupled transformer and, of course, the transformer losses should 
be kept small. 

Appleton, Herd and Watson Watt used a directly heated diode since 
indirectly heated valves did not exist at the time. It is inadvisable to 
use a modern diode because the cathode does not saturate completely 
when the temperature is reduced. The best procedure to adopt is to 
employ a bright-emitter valve having its filament heated by an 
accumulator with a series rheostat for controlling the temperature. 

If a modern diode which is heated from the mains is used and the 
mains potential changes, the diode current may alter slightly and the 
rate of discharge of the condenser will change. Thus, continuous 
fluctuations of the mains potential will prove unpleasant in that the 
time base trace will not recur at constant intervals. Furthermore, if 
the potential to which C 1 is left charged at the end of the pulse of grid 
current is permitted to vary, the repetition rate will vary. The potential 
which C l achieves at the end of the pulse may be dictated by the use 
of a diode whose cathode is connected to the grid end of C l and whose 
anode is connected to some point at, or negative with respect to, the 
potential of the negative supply rail. 

Since the saw-tooth wave-form is obtained by the grid-blocking 
action of the condenser C l5 it is usual to denote this time base and those 
of similar type by the generic term "blocking oscillator." At least six 
methods of application of the blocking oscillator are to be noted : 

(a) The oscillator is employed as the trigger device in a capacitive 
time base and the valve remains in the oscillatory condition only for 
a relatively short period, of the order of a few microseconds. During 
the relatively long trace period, the valve is cut off. The time base 
described above is an example of this type, in which the oscillator may 
be considered as a generator of current pulses. 

(b) The oscillator is used as a generator of brief potential pulses, 






OSCILLATORS AND INDUCTIVE TIME BASES 91 

which may be applied to the grid of another valve to produce current 
pulses. An example of this application is given on page 93 (Holmes, 
Carlson and Tolson's time base). 

(c) The "blocked" period is made shorter than the "on" period. 
In this case, the trace is produced during the oscillator)' period and 
the fly-back occurs when the valve is blocked. For an example of this 
type, see Fig. 96 (pages 126 to 128). 

(d) Blocking oscillators may be triggered from a source of low power 
and, hence, they may be employed to produce a high-power pulse from 
a low-power trigger signal. 

(e) The repetition rate or the pulse duration may be controlled by 
means of a delay line. 

(/) The oscillator may be triggered and used as a frequency divider 
with triggering pulses which arrive at constant intervals or as a counter 
when they arrive at varying time intervals (see also page 257). 

A comprehensive survey of the subject, which includes the design ot 
the necessary transformers, has been made by Benjamin,* and another 
blocking oscillator which may be used as a time base is shown m 
Fig. 66. 

Kobayashi's Time Base 

In 1929 Kobayashif developed a form of blocking oscillator in which 
the blocking action is obtained by a change in the anode potential 
instead of in that of the grid. A valve, having the two windings of a 

transformer in the anode and grid 
circuits respectively, is used to discharge 
a condenser C x which receives its charge 
via a resistance jR,. When the anode 
potential becomes sufficiently high to 
cause anode current to flow, a positive 
pulse is applied via the transformer to 
the grid of the valve so that the anode 
current is cumulatively increased and 
the condenser is rapidly discharged. In 
Fig. 65. Kobayashi's time base Kobayashi's circuit (Fig. 65) a high- 
frequency transformer is used. A later 
suggestion for the use of a low-frequency transformer brought about an 
improvement and, with this change, a modified form of the circuit is 
to be found in many television receivers. This circuit is of little use 
for general-purpose time bases because the available frequency range 
is distinctly limited. 

» R. Benjamin: "Blocking Oscillators," Journ. I.E.E., 1946, 93, Part IIIA, p. H59- 
See also C. L. Faudell and E. L. C. White: British Patent 47J737- 
t M. Kobayashi: U.S. Patent 191 3449- 




0/f.T.+ 



OOut 



GH.T- 









92 TIME BASES 

In Kobayashi's circuit, and in the modified version, the condenser 
C 2 across the cathode bias resistance i? 2 , is made very large in com- 
parison with C l so that a steady bias potential is developed across R z . 
This potential is proportional to the average anode current and tends to 
stabilize the frequency. If, for example, the repetition frequency rises 
as the result of an increase in the H.T. supply potential, the bias 
potential will also increase and will partly counteract the change of 
repetition frequency. 

The synchronizing impulse may be applied by means of a third 
winding on the transformer or from a high impedance circuit to the 
grid of the valve. The latter circuit arrangement must have a high 
value of impedance in order not to reduce the potential applied by the 
transformer to the grid of the valve. A third, and perhaps the best, 
method is to connect a small resistance in series with the earth end 
of the grid winding of the transformer and to apply the synchronizing 
potential across it. 

The greater value of the modified low-frequency transformer circuit 
by comparison with that of Kobayashi is due to the fact that the grid of 
the valve is swung through only one cycle for each discharge of the 
condenser. In this way the condenser is discharged in one operation, 
whereas with Kobayashi's circuit and the original squegger, in which 
the grid is swung positive and negative at a radio frequency, the con- 
denser is discharged in a series of pulses. 

The circuits of Figs. 64 and 65 employ radio-frequency air-core 
transformers while those described later employ iron-core transformers. 

Faudell and White's Time Base 

Faudell and White* have developed a time base which gives a push- 
pull output. It consists of a transformer-coupled blocking oscillator time 
base in which a large inductance is connected in series with the H.T. 
supply to the time base. A condenser is connected between the valve 
side of the inductance and the negative supply rail as shown in Fig. 66. 

The inductances L x and L 2 are of the same value while the capaci- 
tances C 2 and C 3 are of such values that the potentials appearing across 
them are of equal value and opposite phase. The inductance should 
have as high a value of Q as is possible in order that the wave-forms 
appearing at the output terminals may have the highest possible 
potential. This potential may be many times the value of the H.T, 
potential and the ratio 

rail potential 

maximum available potential across L z 

* C. L. Faudell and E. L. C. White : British Patents 479275 and 491934; C. L. Faudell : 
British Patent 4785 11. See also W. S. Percival: British Patent 505252. 



OSCILLATORS AND INDUCTIVE TIME BASES 93 

is proportional to 

maximum current in Cg 

maximum current in L 2 
so that it is advisable to employ for V x a valve which is capable of 
passing a large amount of current and to ensure that the Q values of L 2 
and of the transformer are made as high as possible. In these circum- 
stances, the potential available for charging C 3 may be, say, ten times 
the rail potential and yet the condenser may be allowed to charge only 
to, say, 75 per cent of the rail potential. Thus, if the rail potential is 

1.2 




^WSSb O H.T.+ 

-O Output 



Output 



Control 

Potential 



OH.T.- 



Fig. 66. Faudell and White's time base 



200 volts, the available potential for charging C 3 may be 2,000 volts and 
C 3 may be charged to 150 volts or 7-5 per cent of the total potential 
whereas, had the normal type of arrangement been employed, the 
percentage would be 150 in 200 volts or 75 per cent. This is a high- 
efficiency time base. Faudell and White* also have described a method 
of reducing the duration of the fly-back in a blocking oscillator. 

Holmes, Carlson and Tolson's Time Base 

Holmes, Carlson and Tolson have developed a two-valve time 
base for television reception.! A transformer-coupled oscillator valve 
having a condenser and parallel leak resistance in the grid circuit is 
employed to control a further valve which lies in parallel with a con- 
denser charged through a high resistance. The oscillator circuit is 
conventional except that a low value of series resistance R A is inserted in 
the grid-cathode circuit by means of which a synchronizing potential 
may be impressed across the grid and cathode of the valve. The 
basic circuit is shown in Fig. 67. It should be noted that the con- 
denser and the potential at the secondary terminals of the transformer 
must possess such values that the grid current charges the condenser 
sufficiently to cut the valve off after one half-cycle. Moreover, the 

*C. L. Faudell and E. L.C.White; British Patent 471 737- . 

t R. S. Holmes, W. L. Carlson and W. L. Tolson: "An Experimental Television 
System," Proc. I.R.E., 1934. 22 > P- I2&6 - 






94 



TIME BASES 



charge must hold the valve cut off for a period greater than the time 
required to scan one line of the television raster. This implies 
that Cj should be small enough to permit grid current to charge it, 
during one half-cycle, to a potential which is considerably greater than 
that required to cut off the anode current in V v The value of R t is 
chosen so that C r will discharge to the cut-off bias potential in a time 
slightly greater than the duration of one scanning line. 

This circuit may be considered as consisting of a blocking oscillator- 
pulse generator V x and a trace generator F 2 . 

The condenser C 2 is charged through the high resistance R 3 in series 
with R 2 and is discharged by V 2 when the grid is driven positive by a 



T 



Sec 




-O ft T.+ 



Pri. 



R 3 



Synch. 
O 



-A\ 




£*l£r* 



r\ 



R 



Vr 




c 3 

I 

c 2 



D To 
Amplifier 



Fig. 67. Holmes, Carlson and Tolson's time base 



OH.T.- 



rise in potential on the grid of V v The small resistance R 2 is added to 
provide the required wave-form for producing a saw-tooth current in a 
deflecting coil composed of inductance and resistance in series (the 
inductance L together with the inevitable resistance of the winding). 
This inductance L is composed of two coils placed about the neck of 
the cathode ray tube and is connected in the anode circuit of a power 
amplifier valve which is driven by V % . 

Two such time bases may be employed for television reception. In 
this case the second time base will operate at frame frequency. 

In this, as in all time bases which are capable of operation in the 
absence of a synchronizing signal, the natural frequency is made 
approximately equal to the desired frequency. When a positive syn- 
chronizing signal is impressed across the resistance i? 4 , the grid of V x 
becomes more positive and, provided this signal arrives when the 
charge on C 1 has fallen to such an extent as to permit the grid to rise 
to a potential just beyond cut-off, anode current will commence to 
flow and will rapidly increase due to the cumulative action of the 
valve and transformer. The actual duration of the flow of anode 
current in the valve is determined by the circuit constants and not by 
the duration of the synchronizing signal. It is thus possible to make 






OSCILLATORS AND INDUCTIVE TIME BASES 95 

the valve V t conduct for a period which is either longer or shorter than 
the duration of the synchronizing signal. It is, however, the commence- 
ment of the synchronizing signal which determines the instant at which 
the fly-back commences. 

It will thus be seen that the blocking oscillator part of the circuit, 
i.e. the valve V x and the associated components, perform only one 
essential function, viz. that of keeping the circuit operating in the event 
of the non-arrival of some of the synchronizing pulses due to fading 
in the radio channel. Except for this, the synchronizing signal might 
be applied directly to the grid of F 2 , but, in this case, it would also be 
necessary to ensure that the condenser C 2 could be discharged in the time 
during which the pulse is applied. 

A time base has been developed by Faudell* in which a time base 
charging circuit has its condenser discharged by a triode, the anode 
load of which is the charging resistance. The second valve is also a 
triode having no anode load, the output being taken from a transformer 
in the cathode circuit. The secondary winding of this transformer 
feeds the deflector coils fitted round the neck of a cathode ray tube. 
The cathode of the second valve is back-coupled to the cathode of the 
first via a condenser and a resistance is inserted in the cathode circuit of 
the first valve. 

This arrangement results in improved linearity of the time base trace. 

Bedford f has described an arrangement for improving the wave- 
form in the deflector coil system of a television receiver. 

Faudell and White; have placed the blocking condenser and its dis- 
charging resistance in the cathode circuit of a blocking oscillator in 
order that the fly-back time may be reduced by causing the condenser 
to be charged more rapidly. When the condenser is in the cathode 
circuit it becomes charged not only by the grid current but also by the 
anode current, 

INDUCTIVE TIME BASES 

Blumlein's Electromagnetic Time Base 

This high-efficiency electromagnetic time base was developed by 
Blumlein§ in 1932 and utilizes a hard triode valve and a diode. The 
circuit of the time base is shown in Fig. 68. In this circuit, a constant 
potential is applied to the deflection coil L x (situated about the neck of 
the cathode ray tube) in series with a hard triode valve. A diode is 
connected in the inverse direction across the triode and a condenser C x 
is connected across the inductance L v 

*C. L. Faudell: British Patent 51 1847. fA. V. Bedford: British Patent 401990. 
|C. L. Faudell and E. L. C. White; British Patent 471737. 
§ A. D. Blumlein: British Patent 400976. 



9 6 



TIME BASES 



Neglecting, for the moment, the circuit L 2 C 2i the circuit conditions 
are these : during the forward stroke of the time base wave-form a small 
positive potential is applied to the grid of V lt with the result that the 
anode-cathode impedance is very low. The inductance Lj has a high 

value of — and the impedance of the source of H.T. potential is made 

negligible by the use of a parallel condenser C 3 of high capacitance. 
The rise of current through L 1 is therefore substantially linear. 

Commencing from the moment when the current through L x is zero 
and the cathode ray spot is at the centre of the screen of the cathode 
ray tube; the current in L x rises linearly until a large negative syn- 
chronizing signal is applied to the grid of V v When this happens, the 

Deflector Coi/s 

i-ot^ — nmw — i 

1 v 



Control 3 

Potential a 

c 1 



Fie. 68. Blumlein's time base (electromagnetic) 




0//.7T- 



anode current through V x ceases but the current through L Y at first 
continues and flows into C x thus discharging and recharging it to a 
much higher potential in the reverse direction during one quarter cycle 
of the oscillation of the circuit L t C % . During the next quarter cycle, 
the current through L lt which has already fallen to zero, rises in the 
negative direction while the condenser C\ discharges. When the 
condenser tries to charge once more, however, it can only charge to the 
potential of the H.T. supply, since the cathode of F a becomes negative 
and the current I L flows through the diode instead. The current 
through Lj therefore begins to fall through the action of the diode, 
the controlling pulse ceases and thereafter the cycle repeats itself 
Fig. 69 (a) shows the current I L in L x , Fig. 69 (b) the potential V 6 at the 
grid of V x and Fig. 69 (c) the potential V c across C x (not to earth). 
The following points should be noted : 

(a) The anode potential of V x rises to many (say 20) times 
that of the H.T. supply and these valves must therefore be able 
to withstand the maximum potential. 

(b) Since the anode potential of V x becomes very high, a large 
negative potential must be applied to the grid in order that anode 
current shall be prevented from flowing during the whole of the 
oscillation, or fly-back, period. 



OSCILLATORS AND INDUCTIVE TIME BASES 97 

(c) The negative controlling signal must continue at least until the 
end of the half period of oscillation and must cease before the 
current in L x has fallen from a negative value to zero, at the point 
A in Fig. 69 (a) and (b). The shaded area in Fig. (b) indicates the 
boundaries within which the completion of the controlling pulse 
must be confined. 

The additional tuned circuit L 2 C 2 is tuned to half the scanning 
frequency and is provided to avoid changes in the mean position of the 



(a) Current 
through £ 2 

MWiV- (&) Potential at 
%r X Grid of V z 



(C) Potential 
across Ci 

1 (1.2 &C2 

■ short - 




0- 



k 



/fJ Potential 



circuited) 



Fig. 69. Wave-forms associated with Blumlein's electromagnetic 
time base (Fig. 68) 



time base stroke as seen on the cathode ray tube, 
high value of Q I i 



i.e. 



This circuit has a 
— } where R is the series resistance of the induc- 

R/ 

tance L 2 , and L % is arranged to resonate with C x at a frequency not lower 
than one-third of the scanning frequency. If desired, this circuit may 
be connected in the H.T. supply rail between the condenser C 3 and the 
main tuned circuit L X C V 

White and Blumlein's Time Base 

White and Blumlcin* have developed another high-efficiency time 
base, the circuit of which appears in Fig. 70. In this circuit, a valve 
(not shown) is a blocking oscillator, or other form of saw-tooth generator, 
which produces a positive-going saw-tooth wave-form at the grid of V 2 . 
By correctly adjusting the position of the tapping on the scanning trans- 
former 7\, the potential appearing at the junction of L 2 and i? 5 , while 
anode current flows in F 2 , can be made almost equal to that of the H.T. 
supply. In this case, little or no current will flow in the resistance R B 
and the anode current of V 2 will pass through the diode F 3 . 

When the anode current in V 2 is cut off by the saw-tooth input 

*E. L. C. White and A. D. Blumlcin: British Patent 482370; E. L. C. White: 
British Patent 529383; O. H. Schade: British Patent 566835- 






9 8 



TIME BASES 



potential on its grid, the current flowing in 7\ decreases to zero while 
executing a quarter cycle at a high frequency whose value is determined 
by the stray capacitance across the transformer. The current then 
increases in the negative direction through the diode V% and the circuit 
C 5J R 5 as shown at A in Fig. 71. Current therefore flows in F 3 for 



To Deflector Coils 



T 2 



L 3 



Rs 



12 






n 



oh 



t-VWW- WAAr- 



-OH.T.+ 




Fig. 70. White and Blumlein's time base 

almost the complete cycle and in V 2 for a lesser period and the average 
diode current exceeds the value of the anode current in F 2 . As already 
stated very little current flows in R h and the condenser C 5 becomes 
charged in the sense shown in Fig. 70 and so stores energy which 
is added to that of the H.T. supply and provides the necessary power 
for overcoming the back-E.M.F. of L 2 . 




Fig. 71. Wave-forms pertaining to the circuit of Fig. 70 

The early part of the saw-tooth is produced by the decrease of current 
in L 2 at a frequency determined by the value of C 5 . This current, 
which flows through V 3 , is shown at A' in Fig. 71. The second part of 
the saw-tooth is produced by the increase of current in L x and L 2 
flowing through V 2 (B in Fig. 71) and through V 3 (A" in Fig. 71), the 



OSCILLATORS AND INDUCTIVE TIME BASES 99 

two curves adding together to produce the full line curve C in the 
same figure. In order to obtain a smooth change-over when the curve 
C passes through zero during the forward stroke of the saw-tooth 
current wave-form, it is necessary to ensure that the leakage inductance 
of the windings in L x and L 2 of the transformer 7\ shall be as small as 
possible. 

In order to avoid distortion of the wave-form of the anode current 
in V 2 (curve B in Fig. 71) when the amplitude is changed, the control 
of the scanning amplitude is carried out by adjustment of the resistance 
R Q , which simultaneously increases the potentials at the screen-grid of 
V 2 and the anode of V lf and hence, of the amplitude of the saw-tooth 
potential applied to the grid of the valve V 2 . 



CHAPTER VII 

POLAR CO-ORDINATE, MULTIPLE AND VELOCITY 
MODULATION TIME BASES 



POLAR CO-ORDINATE TIME BASES 
The art of plotting a wave-form in polar co-ordinates consists of the 
fixation of a number of points in one plane, the radial distance of each 
point from the centre representing the potential (or any other quantity) 
existing at the instant of time indicated by the angle between this 
radius and a reference axis. In using a circular time base trace, it is 
usual to employ a constant angular velocity, time being indicated by 
the angular position while the radius of the circle is controlled by the 
potential or current under examination. This method of delineation 
offers certain advantages over the ordinary Cartesian co-ordinate time 
base, since it provides a longer trace and avoids the necessity for a 
fly-back. Moreover, the constant velocity is produced by a pure 
sinusoidal wave-form, which is readily available, instead of requiring 
a saw-tooth wave-form. Nevertheless, the polar co-ordinate time base 
remains unpopular for ordinary oscillograph work largely because we 
are more familiar with the use of squared paper than with the polar 
variety. In cases where the significance of a polar diagram is readily 
understood, the circular time base proves extremely useful. 

Lilienf eld's Circular Time Base 

Lilienfeld's time base has already been described (see page n), but 
the method by which the signal to be examined is applied to the cathode 
ray tube is of interest. Since all the deflector 
plates are already employed for the time base 
they are not easily made available also for the 
application of the signal, although such a pro- 
cedure is possible. Moreover, if the signal 
were to be applied between only one pair of 
the deflector plates, a polar oscillogram would 
not be obtained since the deflection would 
not always be at right angles to the time 
base trace, but would appear as in Fig. 72. 
The signal potential is usually made to appear 
in series with the potential which is applied 
to the final anode of the cathode ray tube. 
In this way, the final anode potential is modulated and, hence, the 
deflection sensitivity of the tube is altered in accordance with the 




Fig. 72. 
Effect of applying a signal 
in a vertical direction when 
using a circular time trace 



POLAR CO-ORDINATE TIME BASES 



101 



instantaneous value of the resultant potential since the beam velocity 
is caused to change (see Appendix I). This method is fairly satisfactory 
for relatively small deflections when a gas-focused tube is employed, 
but the ionizing conditions are altered and the maximum amount of 
the deflection is a compromise between obtaining a reasonable size of 
image and the reduction of the resultant defocusing effect to avoid 
undue blurring. This method is not so easily applied to high- vacuum 
electron-optically focused tubes because it is necessary to apply a 
fraction of the signal to the focusing anode. When a high-vacuum 
tube is employed without this refinement, even with small deflections, 
the electron-optical focusing system is disturbed because the focusing 
potential gradients are altered by the change in the final anode potential 
and the method is thus seen to be somewhat unsatisfactory. 

A serious disadvantage of Lilienfeld's circuit, which applies also to 
Dye's time base (see below), is that the time base potentials are not 
applied in push-pull. This omission results in distortion of the 
circular trace unless the cathode ray tube is specially designed to 
reduce trapezium distortion to a minimum (see pages 262 and 276). 

If the source of supply for the time base is impure, i.e. when har- 
monics are present, circular time bases are often unsatisfactory since 
some irregularity in the velocity and circularity of the trace is generally 
unavoidable. The irregularity is introduced because, so far as the 
harmonics are concerned, the impedance of the condenser is reduced 
while that of the resistance remains unchanged. 

Dye's Circular Time Base 

Dye* has developed a method of avoiding most of the errors due to 
impurities in the wave-form of the supply potential. In Dye's circuit, 
a modification of which is shown in Fig. 73,! the resistance employed in 
Lilienfeld's circuit is replaced by a parallel tuned circuit L X C X . The 
resonant impedance of this circuit, which is tuned to the frequency of 

the supply, is equivalent to a pure resistance of value ■* , where 

R x is the effective resistance of the inductance L x but, at the harmonic 
frequencies, the inductance has relatively little effect and the circuit 
behaves as a condenser. The impedance of the tuned circuit falls 
at the harmonic frequencies in the same way as does that of the 
condenser C 2 , with the result that the amplitudes of the harmonic 
potentials applied to the two pairs of deflector plates are substantially 
reduced and the circularity of the trace is not seriously impaired. I n 

* D. W. Dye : " Improved Cathode Ray Tube Method for the Harmonic Comparison 
of Frequencies," Proc, Phys. Society, 1925, 37, p. 158. 

t In Dye's original circuit, the resistance J?! is placed in the capacitive arm of 
the tuned circuit. 



102 



general, the impedance 



TIME BASES 

1 of the tuned circuit at the fundamental 
frequency is made equal to and should not be too high since, 

o>C 2 

otherwise, a very small value of capacitance is required. Moreover, in 
this case, the circuit impedances may be sufficiently high to pick up 
unwanted interference due to the presence of stray capacitance between 



I 



C z 





Ciwmm 










'"U 


54 

























1 — i 


— ^ 


T/'na/ Anode 



Fig. 73. Dye's improved circular time base 

the deflector plates (and their associated wiring) and other circuits in 
the vicinity. 

Since, in order to produce a circular trace, the two arms of the net- 
work must have equal impedances if the deflection sensitivities are 
equal, we may say that 

and if C L equals C 2 , then 

i 

or, when the sensitivities are unequal, 

m x wCj 

where m 1 and m t are the sensitivities, in millimeters per volt, of the 
pairs of plates to which the tuned circuit and the condenser C 2 are 
respectively connected. 

Thus, by making — -±- smaller than -— , e.g. by increasing the value 

off M v compensation for unequal sensitivities of the two pairs of deflector 
plates is easily attained. In this case, the tuned circuit is connected 
across the pair of plates which has the higher sensitivity. Fig. 74 shows 
the appropriate circle diagram. 






POLAR CO-ORDINATE TIME BASES 



103 



It should be noted that, if the alternating potential source is of low 
impedance, it is helpful to insert in the feed line to the input network 
a series resistance across which the harmonics can develop a potential. 




Fig. 74. Circle diagram for Dye's time base 

Watson Watt, Herd and Bainb ridge -Bell's Circular Time Base 

In 1932, Watson Watt, Herd and Bainbridge-Bell* described a 
circular time base which they developed at the Radio Research Station 



0//.Z+ 




Q Earth 

Signal 
Fig. 75. Watson Watt, Herd and Bainbridge-Bell circular time base 

at Slough. The circuit of this time base is shown in Fig. 75. A phase- 
splitting circuit C X R X is employed, as in Lilienfeld's circuit, to provide 
two phases which are 90 out of phase. The value of R t is made equal 
t -L. The potentials across R x and C x are fed to the primary wind- 
ings of two transformers having their secondary windings centre-tapped. 

• R. A. Watson Watt, J- F. Herd and L. H. Bainbridge-Bell: The Cathode Ray 
Tube in Radio Research, published by H.M. Stationery Office. 



104 



TIME BASES 



The result is to provide a four-phase supply to the grids of the 
four valves, and the amplified potentials are thereafter applied to the 
four deflector plates of a cathode ray tube. The potential to be examined 
is applied to the grids of all four valves in series with the necessary 
bias potential. Since each pair of plates is operated in push-pull, the 
signal does not itself produce a deflection of the spot, but modulates 
the four-phase input and thus controls the diameter of the circular trace. 
This circuit will function more satisfactorily if variable-mu valves are 
employed. These valves had only just been developed when the circuit 
was originally described but were not available when the circuit was 
developed. 

The Polar Co-ordinate Oscillograph 

A polar co-ordinate oscillograph usually employs a special form of 
cathode ray tube or other oscillographic device which has been specially 
designed for the purpose of depicting oscillograms in polar co-ordinates. 

Von Ardenne,* and Dowling and Bullenf have independently 
developed polar co-ordinate cathode ray oscillograph tubes. Von 
Ardenne's tube and the circuit employed with it are shown in Fig. 76. 



Wave Fortn to be examined 







FlO. 76. Von Ardenne's polar co-ordinate oscillograph 

The tube contains two concentric cone-shaped deflectors across which 
the potential to be examined is connected. The resultant form of the 
image is shown in Fig. 77. This form of cathode ray tube has the 
advantage that the final anode potential remains fixed and it is, there- 
fore, possible to obtain larger deflections without defocusing than is the 
case when the signal is applied as a modulation of the final anode 
potential. 

Von Ardenne has also employed electromagnetic deflection methods 
for this purpose. 

* M. von Ardenne: "A New Polar Co-ordinate Cathode Ray Oscillograph with 
extremely Linear Time Scale," Wireless Engineer, 1937, 14, p. 5. 

•f J. J. Dowling and T. G. Bullcn : " Precision Measurements with a Radial Deflection 
Cathode Ray Oscillograph," Proc. of the Royal Irish Academy, Section A, 1937, 44, p. 1. 






POLAR CO-ORDINATE TIME BASES 



10: 



The form of the image, as shown in Fig. 77, is not in true polar 
co-ordinates in the mathematical sense of the term, but it is difficult 
to find a name which truly describes the arrangement. 

MULTIPLE TIME BASES 
The term multiple time base may be applied to those systems in 
which two time axes, generally at right angles to one another, are 
provided. Multiple time bases have been used for many purposes, and 




Fig. 77. Polar co-ordinate oscillogram obtained with Von Ardenne's 
polar co-ordinate oscillograph 

their use is a necessity where it is required to examine potential or other 
changes of short time duration having a large and variable time separa- 
tion, or where the potential changes recur only after long intervals. Two 
separate time bases are connected to the cathode ray tube so that one 
may provide horizontal, and the other vertical, deflection. This is pre- 
cisely the condition which arises in television transmission and reception, 
in which case linear time bases are required. Multiple time bases may, 
however, be linear, sinusoidal, spiral or radial, and other more complex 
forms may be envisaged. 

One of the two time bases forming the multiple time base may be 
used to lock or synchronize the frequency of the other with its own 
frequency so that they remain in step. 

The Spiral Time Base 

If a linear saw-tooth time base potential is employed to control the 
amplitude of a circular time base, a spiral or a radial time base deflection 



io6 



TIME BASES 



results. When the period of the linear time base potential is « times 
the period of the circular trace, a stationary spiral trace appears, where 
n is any integer. When the linear trace period differs slightly from n 
times the circular trace period, the spiral rotates. 

Go deck's Radial Time Base 

If the frequency of the circular time base is reduced until n is a 
small fraction of unity, the spiral trace develops into a radial one. 

The radial time base is used in the P.P.I, (plan-position indicator) 
which forms a part of some types of radar equipment. In general, 
however, electromagnetic deflection is employed and the radial deflec- 
tion is produced by currents of appropriate wave-form which are caused 
to pass through coils which are then rotated at a constant rate round the 
neck of the cathode ray tube in order to cause the radial saw-tooth to 
appear to rotate. In other cases, the coils are fixed and out- of- phase 
currents of appropriate wave-form are passed through the two pairs of 
coils in order to cause the radial time base to rotate. Williams, Howell 
and Briggs* have published information on this subject. 

Godeckf has developed a radial time base, the circuit of which is 
shown in Fig. 78. Obviously, by altering the ratio of the two supply 
frequencies, this time base will also function as a spiral time base. 

By means of a phase-shifting network C 2 C 3 R 2 R 3 , in which 

R Z =R Z = = 

(oC 2 6>C 3 

a four- phase supply is obtained and employed to provide circular 
deflection of the beam. The four phases are connected to the mixer 
grids of four hexode valves, each of which drives one of the four 
deflector plates of the cathode ray tube. It will be appreciated that the 
method of obtaining the four phases, although circuitally different from 
that of Fig. 75 is essentially the same; in this case the transformer 
precedes the phase-shifting circuit. 

The time base supplying the linear deflection in a radial direction 
on the tube screen is applied, in the same phase, to the signal grids of 
all four valves and serves to modulate the circular time base deflection. 

The circularity of the trace is controlled, with the anode load potentio- 
meters, i? 4 to R 7 inclusive, set at an equal value, by adjustment of the 
variable resistances R 2 and R 3 of the phase-shifting network. These 
two resistances should, in theory, be of equal value, but small adjust- 
ments will be necessary to compensate for any inequality in the value 

*F. C. Williams, W. D. Howell and E. H. Briggs: "Plan-position Indicator Cir- 
cuits, " Journ, I.E.E., 1946, 93, Part II1A, p. 1219. 

t J. E. Godeck: Royal Aircraft Establishment, Famborough, late Research Depart- 
ment. A. C. Cossor, Ltd., London; G. Gobau: "Eine Methode zur Radialen 
Ablenkung an der Braunschen Rohre" ("A method of radial deflection in a Cathode 
Ray Tube"), Hockfrequenztechnik, 1932, 40/1, p. 1. 



MULTIPLE TIME BASES 



10* 



of the capacitances C 2 and C 3 . The mean diameter of the circle is 
controlled by adjustment of the input potential to the phase-shifting 
device by means of the potentiometer /?„ while the circularity may be 
given a final adjustment by control of the anode load potentiometers. 
All these controls, except R v are normally preset. 

The degree of modulation by the linear trace and, hence, the area of 
the screen covered by the image may be adjusted by varying the setting 




Modulation 
from r 7 
Linear 
Time Base 




-O/f.T- 



FiG. 78. Godeck's radial time base 

of the variable resistances, i? 8 to R u inclusive, which are in series with 
the four 50-microfarad electrolytic condensers. These resistances and 
condensers are, together, connected between the cathode of each valve 
and the negative supply rail The resistances may be ganged if they 
are first matched and if they obey the same law. 

This form of time base is extremely useful where it is desired to 
produce a long trace. For a linear time base repetition frequency of 
200 times that of the circular trace and a linear trace length of 5 inches, 
which is easily attainable on a 12-inch tube, the total trace length is 
r.ooo inches. Neglecting the fly-back time, if the circular time base 
frequency is 50 cycles per second, 1 inch of trace represents a time of 
20 microseconds. In order to avoid lost time, the fly-back velocity 
should obviously be of a high order and, in the interest of image clarity, 



ioS 



TIME BASES 



VELOCITY-MODULATION TIME BASE 



109 



:> 



the fly-back should be blacked-out by application of the fly-back 
signal, via a differentiating circuit, to the grid or cathode of the cathode 
ray tube according to the polarity of the signal. 

With this type of time base, the signal to be examined is usually 
applied to the grid or cathode of the oscillograph tube so that it 
modulates the brightness of the trace. 

VELOCITY-MODULATION TIME BASE 
Most of the linear time bases described in this book may be used 
for television reception but, obviously, the most suitable are those which 
have the requisite degree of linearity. For velocity-modulation systems 
of television, however, it is essential to employ a velocity- modulated 
time base at least for the line deflection.* Although it is unlikely that 
velocity-modulated television systems will be employed in the future 
it is more than probable that occasions will arise in which a velocity- 
modulated time base will be required. Such a time base differs from 
an unmodulated one only in that the charging current is modulated in 
a known manner. Thus, for a television time base, the charge upon 
the condenser at any instant is dependent upon the total amount of 
illumination which has fallen upon the photo-electric cell from the 
beginning of the trace until the instant under consideration. Moreover, 
the instantaneous rate of charge of the condenser is dependent upon 
the instantaneous illumination upon the cell. The charge upon the 
condenser is thus dependent upon past, as well as upon present, 
conditions of illumination. 

This result is achieved by the application of the amplified photo-cell 
output potential to the grid of a charging pentode. The circuit arrange- 
ment is shown in Fig. 79, in which the resistance R t is normally zero. 
In this form of television transmission, film scanning is more usually 
employed and a cathode ray tube is used at the transmitting end. This 
tube provides a raster which scans the film, a lens being used to focus the 
light from the raster upon one frame of the film, and the light, after 
passing through the film, falls upon the photo-cell. As a result of the 
application of the amplified and integrated (by means of C 2 ) output of 
the photo-cell to the X deflector plate of the cathode ray tube, the 
velocity of the scanning spot (writing speed) is modulated in accordance 
with the variations of film transparency and an image of the picture on 
the film appears on the screen of the transmitting cathode ray tube. In 
essence, therefore, it is only necessary to transmit the two deflector plate 
potentials to a distant cathode ray tube in order to reconstruct the 
transmitted image since all the necessary information regarding both 

• L. H. Bedford and O. S. Puckle: "A Velocity-Modulation Television System," 
Journ. I.E.E., 1934, 75, p. 63; British Patent 427625. 



the instantaneous brightness and the position of each element of the 
transmitted picture is implicit in the signal. This is quite different 
from the intensity-modulation system of television transmission in 
which the signal contains only information regarding the instantaneous 



Discharge 




oN.T.i- 



To radio 
-O transmitter. 
To an X plate 

O ofthe scanning 

Input from >*-4 \* cathode ray 

Photo-ceil j S 7 i tube - 

o 1 T 1 — I oz/.z"- 

Fig. 79- Velocity-modulation time base arranged to com- 
bine the advantages of velocity- and intensity-modulation 

brightness of each picture element except at the beginning of each 
scanning line, when a positional signal is transmitted. 

If the resistance R x in Fig. 78 has a finite value, it enables the circuit to 
combine the advantages of velocity- and intensity-modulation, since 
the additional potential appearing across the resistance is proportional 
to the instantaneous value of the charging current and, hence, to the 
instantaneous value of the illumination. As will be seen from Fig. 78, 



-QH.T.+ 



R, 



from /("actio 
Receiver 

O 

Fig. 80 



C\ 




-^OToXP/ate 



B 7° 

-=— O Modulating 

Grid 



—OHX- 



Basic circuit for the reception of combined velocity- and 
intensitv-television transmissions 



an X plate of the scanning cathode ray tube is connected to a point in 
the resistance R v 

At the receiver the basic circuit of Fig. 80 is employed. The signal 
applied to the grid of V x is a copy of the signal appearing at the anode 
of V x in Fig. 79. If the time constant C 1 R l is made equal to that of 
C 1 R 1 in Fig. 79, and if R 2 is made vanishingly small, the potential 
appearing at the point A is proportional to the deflection of the scanning 



no 



TIME BASKS 



VELOCITY-MODULATION TIME BASE 



in 



A C 

1 




B D 
(a-) 



Time 



beam at the transmitter, while that at the point B is proportional to the 
time derivative of the potential at A. Hence, these two potentials are 
suitable (after appropriate amplification) for application to the deflector 
plates and the modulating grid respectively of the receiving cathode 
ray tube. That this is so may be seen from Fig. 81 in which (a) is an 

imaginary white rectangle 
in a dark field which it is 
desired to transmit. In this 
diagram AB and CD re- 
present a pair of consecu- 
tive scanning lines, the line 
BC, which is not neces- 
sarily straight, representing 
the fly-back. In Fig. 81 (b) t 
the corresponding deflector 
plate potential on the film- 
scanning cathode ray tube 
is plotted against time while 
at (c) is plotted, to the same 
scale, the time derivative of 
the potential at (b). In 
Fig. 8 1 (b) the potential 
corresponding to the scan- 
ning line AB falls steeply at 
first while the spot scans 
rapidly over the dark field, 
then more slowly as the spot 
moves over the white rect- 
angle and again more 



j u 



\ 



(c) 



U L 



Fig. 8 1 .Wave-forms 
obtained (b) with a 
velocity - modulated 
television transmit- 
ting system (d)with 
the addition of the 
intensity modula- 
tion (c) 




rapidly over the second portion of the dark field. There next follows 
a rapid rise of potential during the fly-back but, since the discharge valve 
is in control of the circuit at this period, the transparency variations in 
the film have no effect on the velocity of the trace. The derived curve (c) 
is similar to that which would appear if the film were being scanned 
for the intensity system of television transmission except that the time 
scale is opened in the light portions and closed in the dark parts of the 
field. The curve shown in Fig. 8i (d) is the sum of the curves of (b) 
and (c) and is the actual wave -form transmitted in the velocity plus 
intensity system. Since the curve (d) is the sum of two curves of which 
one is the time derivative of the other, the potentials of the two com- 
ponent curves can be separated again by means of the circuit of Fig. 8o, 
provided that the conditions already specified are complied with. 
If the output from the point B in Fig. 8o is fed to the grid of the 






valve V x of a further time base similar to that of Fig. 79 (but with R x 
removed), the arrangement can be employed to produce the frame 
scanning signal at the receiver. The pulses appearing at the point B 
at each line fly-back are fed, via a short-time-constant differen- 
tiating circuit, to the grid of V x so that an increment of charge is fed 
into the condenser in the anode circuit at each line fly-back. This 
results in each successive line at the receiver being made to occur in a 
position adjacent to the preceding line. For obvious reasons the process 
is known as "ratcheting." Note that the valve V t in Fig. 79 behaves 
as an integrator. 

Since the velocity-modulation television system described transmits 
a wave-form which is represented by the variations in transparency of 
a cinema film, the system may obviously be employed to generate any 
desired wave-form by placing the appropriate transparency between 
the cathode ray tube and lens system and the photo cell. The wave-form 
is then available for external use (see also page 363). 

The circuit of Fig. 79, with R x adjusted to zero, may be employed 
to increase the velocity of, and thus to expand, a time base trace during 
the time when a narrow pulse appears upon the cathode ray screen. 
In order to achieve this effect, the pulse is applied in a positive direction 
to the grid of the charging valve as well as to the Y deflector plates of 
the tube. As a result, the charging current is increased during the 
pulse. It is, of course, necessary to bias the grid of the valve V x so as 
to reduce the charging current in the absence of a pulse. 



CHAPTER VIII 
LINEARIZATION OF THE TRACE 

The Meaning of Linearity 

Whkn it is stated that a saw-tooth wave-form is non-linear by a 
certain percentage, it is to be understood that it is the slope of the 
wave-form which varies by this percentage. The error which will 
result when the wave-form is applied to an oscillograph is, however, 
not to be confused with the variation of the slope of the wave-form, 
which results in a variation of the velocity of the spot. If the trace were 
truly linear, it would conform to the equation x—kt, where x is the 
deflection of the cathode ray spot or the amplitude of the current or 
potential which produces it, k is a constant and t is the time from the 
commencement of the trace. 

In order to measure time intervals with the aid of a nearly linear 
time base, the simplest method, though not the most precise, is to 
place before the screen a squared celluloid scale marked in centimetres 
and millimetres. The amplitude of the sweep may then be adjusted 
so that the trace exactly spans a whole number of divisions, say 10 or 
20 centimetres, so that any time interval may be measured from the 
beginning of the trace. Neglecting the possibility of error in deter- 
mining the duration of the trace, it is obvious that the non-linearity 
of the trace will not introduce any error at the two ends of the scale since 
the trace has been adjusted to a known length. One must, however, 
expect that measurements made in the region of the middle of the scale 
will be inaccurate if the time base is not perfectly linear. 

The error due to the non-linearity of the sweep may be described 
as a displacement error to avoid confusion with the velocity error. 
Usually, the average velocity of the spot is higher during the first half 
of the trace than during the second half and consequently the spot is 
displaced relative to the scale by an amount which reaches a maximum 
near the centre of the trace. The maximum value of the displacement 
error is always much less than the maximum variation in velocity, as 
will be seen from Fig. 82, which has been plotted for the ordinary 
exponential charging characteristic. 

When a condenser is being charged through a resistance, the shape 
of the charging curve and the resulting error are determined by the 
maximum condenser potential expressed as a percentage of the supply 
potential, i.e. the percentage charge. In Fig. 82 the displacement error 
has been plotted against the percentage charge and the curve may be 
applied to any time base in which the charging current decays 






LINEARIZATION OF THE TRACE 






113 



exponentially. Even a time base employing a nominally constant- 
current charging device generally falls within this category even if the 
effective supply potential is considered instead of the actual potential. 
(No precisely constant-current time base charging circuit has yet been, 
or is likely to be, realized.) Thus, if the variation in charging-current 
is known, the resulting displacement error may be obtained from the 
figure. 

Since the velocity of the spot at any instant is proportional to the 
instantaneous value of the charging current, provided no distortion is 



8L 



3 






.1 



S 



. 1 I 

+ 



7 2 3 4 5 10 20 30 40 50 100 

% fall in spot velocity at end of trace or, 
% Fall in charging current at end of trace or, 
% Charge at end of trace 

Fig. 82. The displacement error in a time base trace 

introduced as, for example, by an amplifying stage, the percentage fall 
in velocity is identical with the percentage fall in current. Furthermore, 
the percentage fall in the charging-current is equal to the percentage 
charge. The horizontal scale of Fig. 82 may thus be read as percentage 
fall in spot velocity, percentage fall in charging-current or percentage 
charge. 

It is notable that an 8 per cent fall in current produces a maximum 
displacement error of only 1 per cent. In the reproduction of a tele- 
vision image this amount of distortion would be slightly objectionable. 



II 4 



TIME BASES 



LINEARIZATION OF THE TRACE 



«S 



Linearization of the Trace 

For the majority of time bases required for general purpose investiga- 
tions, the degree of linearity obtainable with a pentode charging valve 
is usually sufficient, but for high-quality television receivers and for 
some other purposes, including precision time-measuring circuits for 
radar, an improved degree of linearity is required. A number of 
methods are available, and these generally employ one or other of the 
following devices for obtaining the required compensation: 

(a) The inverse curvature of a valve characteristic (see below). 

(b) An auxiliary time constant circuit (page 117). 

(c) A variable inductance (page 124). 

(d) Feed-back arrangements (page 128). 

Particularly useful papers on circuits and methods of linearization of 
the trace have been written by Williams and Moody, Pout and Clarke.* 

(a) LINEARIZATION BY MEANS OF INVERSE 
CURVATURE 

A number of methods have been developed which compensate to a 
certain degree for the curvature of the charging characteristic by 
inserting a device having a similar but inverse characteristic. An 
arrangement which is often adopted is to amplify the potential across 
the condenser by means of a valve in such a way that the valve curva- 
ture is employed as a linearizing means. It is fortunate, in this respect, 
that the curvature of a part of the grid volts-anode current characteristic 
of a valve approximates to the inverse of an exponential characteristic. 

Bedford and Stevens' Method of Linearization 

In order to meet the cheapness requirements of commercial tele- 
vision competition, Bedford and Stevens f avoided the use of a pentode 
charging valve and reverted to a charging resistance which, of course, 
resulted in an exponential output. It should be noted in passing that, 
provided the condenser is charged to only about 7 per cent of the 
supply potential, the output is sufficiently linear to be used with the 
average cheap television receiver, but, even so, there remains a certain 
amount of distortion of the picture due to non-linearity. In this case, 
of course, the output amplitude is very low, unless an extremely high 
charging potential of the order of several thousand volts is employed. 

In the Bedford and Stevens circuit a reasonable charging potential 
is used and the charging characteristic is deliberately allowed to be 

* F. C. Williams and N. F. Moody: "Ranging Circuits, Linear Time Base Genera- 
tors and Associated Circuits," Journ. I.E.E., 1946, 93, Part IIIA, p. 1 188; H. W. Pout: 
"Precision Ranging Systems for Close-range Weapons," Journ. I.E.E., 1946, 93, 
Part IIIA, p. 380; A. C. Clarke: "Linearity Circuits," Wireless Engineer, 1944, 21, 
p. 256. 

f L. H. Bedford and W. H. Stevens: British Patent 474623. 






exponential, the trace being linearized by means of a valve amplifier 
having a suitable characteristic. If the amplifier characteristic is so 
shaped, the input so proportioned and the anode load of such a value 
that, at each point on the dynamic characteristic, the anode potential 
is equal to the difference between the charge on the condenser and the 
potential required to produce a linear characteristic, the non-linearity 
of the condenser charge may be compensated. This is shown in Fig. 83 . 
The charge on the condenser is applied to one deflector plate of the 
cathode ray tube, and a suitably scaled-down copy of this potential is 

1500 >v 




Voltage required to be 
produced at the atode 
of the valve 



550 V 



SOV 



Anode Current 



Time *- 






Fig. 83. Curves showing the relationships existing in the Bedford and Stevens 
method of linearization 

fed to the grid of the amplifier. The other deflector plate is connected 
to the anode of the amplifier and the result is a fairly linear sweep 
although the mean potential does not remain constant since the 
potentials applied to the two deflector plates are not the same at every 
instant. Therefore the method functions by generating one signal 
whose errors cancel those of another. This circuit results in a material 
reduction of cost. 

Bedford's Method of Linearization 

When using one of the author's time bases with the addition of an 
amplifying stage, Bedford developed a circuit for controlling the degree 
of linearity. This arrangement is shown in Fig. 84. A cathode potentio- 
meter R % is inserted in the amplifier circuit and a connection is taken 
from the moving contact of the potentiometer to the cathode of the 
charging pentode. When the potential on the grid of the amplifier is 



n6 



TIME BASES 



near zero, i.e. when the anode potential of V t is at its highest value, 
maximum current flows in the amplifier cathode circuit and the cathode 
of the charging valve is made more positive, thus reducing its anode 
current. Conversely, when the grid is negative, the cathode of the 
charging valve is likewise made less positive, resulting in an increased 
charging current. The effect produced is shown in the characteristic 
curve for the pentode charging valve which appears beside the circuit 
diagram. It will be apparent that, by altering the setting of the potentio- 
meter, the charging curve of the valve V x may be altered so that the 
effective slope impedance approaches infinity (horizontal) and hence 
the imperfection in the linearity can be made to approximate to zero. 



0//.T+ 




/{node Volts 

Excursion 

ofVt 




Resultant Current 
after Linearization. 
Characteristic 
of Valve V\ 



oSynch. 



Fig. 84. Bedford's method of linearization 

Jenkins' Method of Linearization 

Jenkins* has developed a method of linearizing a time base: the 
circuit is shown in Fig. 85. In this time base the output potential is 
amplified by a second valve V 2 and the curvature of the condenser charg- 
ing characteristic is modified until it is such that it can be compensated 
by the inverse curvature of the amplifying valve. This aim is achieved 
by utilizing a potentiometer as the anode load of the discharging valve 
V 1 and by connecting a condenser C 2 , which is of sufficient capacitance 
to hold the potential across it steady, between the slider and earth. 
The time base condenser C l is connected across the discharging valve 
and is held substantially discharged except when an externally generated 
square- wave negative pulse of predetermined time duration is fed to the 
grid. Such a pulse determines the duration of the charging time and it is 
a condition on which the scheme is based that this time be substantially 

t 
fixed or, alternatively, that -^ remains nearly constant, where t is the 

pulse duration or charging time and C x the capacitance. 

* J. W. Jenkins: Research Department, A. C. Cossor, Ltd., London. 



LINEARIZATION OF THE TRACE 



117 









The primary effect of sliding the condenser C 2 along the potentio- 
meter is to alter the charging potential E and the charging resistance R 
while keeping the ratio EjR substantially constant. EjR remains 
substantially constant because t is short compared with the time base 
repetition interval and therefore the condenser C 2 has a charging time 
which is short compared with the discharging time. Since EjR 

-0//.ZV 



OOut 



is 




Control 
Potential 



0//.7T- 



Fig. 85. Jenkins' method of linearization 



constant, C t R is directly proportional to E while the anode current in 
V 1 is cut off, and therefore the shape of the charging curve depends on 
the position of the potentiometer slider. By a suitable choice of the 
position of the slider, the slope of the charging curve may be adjusted 
until it becomes the inverse of the characteristic of the amplifying valve 
F 2 . The start of the trace is substantially constant in slope, as can be 

E 

seen from Table II (page 118), since ~ - R is constant. 

The lower the value of E the more curved does the charging 
characteristic become. 

(b) LINEARIZATION BY MEANS OF AN AUXILIARY 
TIME CONSTANT CIRCUIT 

Hawkins* Method of Linearization 

Hawkins* has developed a method of modifying an exponential wave- 
form of potential across the condenser of a time base by the addition 
of an integrating circuit. The circuit arrangement is shown in Fig. 86, 
from which it will be seen that the time base condenser is in two 
portions, C x and C 2 , which are charged together (in series) through the 
resistance R x and discharged through a thyratron which is indicated in 
the figure by a switch. 

* G. F. Hawkins: British Patent 511600: "Linear Saw-Tooth Oscillator," 
Wireless World, 1939, 44, p. 42s. See also Appendix IV, p. 323. 



xi8 



TIME BASES 



Table II 



Effect of E on rate of rise of potential across C 1 when the 

E 

CURRENT -5 IS HELD CONSTANT. 
R 



E (volts) 



C\i? (seconds) 



280 



140 



o-oi 



70 



0-005 



35 



0-0025 



Time in micro- 
seconds from 
the commence- 
ment of the 
pulse or sweep. 


Potential across Cj (volts) 








Poten- 
tial 


Per cent 
Charge 


Poten- 
tial 


Per cent 
Charge 


Poten- 
tial 


Per cent 
Charge 


Poten- 
tial 


Per cent 
Charge 


100 


1-40 


0-50 


1-40 


1*00 


1-40 


2-oo 


1-36 


3-90 


250 


3-SO 


1-25 


3-5° 


2-50 


3'43 


4'9° 


3*33 


950 


500 


7-00 


2-50 


6-86 


4-90 


665 


9*5° 


6-33 


i8'i 


1000 


i3'7 


4Q0 


13*3 


9-5° 


12*6 


180 


12-55 


35-8 



OOut 



Note that if the rate of charge is linear, the percentage charge will increase in the 
same ratio as the ratio of the times quoted in the left-hand column. 

Assume that the condensers C x and C 2 have just been charged. The 
condenser C 8 will then be charged to a potential which is less than 

that of C 2 because the potential across 
Cj is the source from which C 3 charges 
through R % . For this reason, C 3 cannot 
reach the potential of C 1 unless the latter 
remains at a constant potential for a time 
which is long compared with the value 
of the time constant C 3 R 2 . 

The switch is now closed and the 
potentials across C x and C 2 commence to 
fall while the potential across C 3 falls at 
a lower rate due to the large time con- 
stant C Z R Z . C x and C 2 discharge rapidly 
through the switch. In practice, C 2 and C 3 are of the same order of 
capacitance and R 2 is of the order of 100,000 ohms. 

When Cj and C 2 are discharged and the switch is again opened, 
C 3 still retains a considerable charge. C l and C 2 now start to charge 
once more, but C 2 charges from two sources, from the H.T. supply 
through Rj_ and from C 3 through R 2 so that C 3 continues to discharge 




Fig. 86. 



Hawkins* method of 
] inearization 






LINEARIZATION OF THE TRACE 



119 



until the potentials of C x and C 3 are equal. After this C 3 charges 
from Cj. 

As a result the potential wave-form across C x and C 2 is exponential, 
while that across C 3 is approximately parabolic. The output is taken 
across C 2 and C 3 in series and, if the capacitances of C 2 and C 3 are 
suitably chosen, the resultant wave-form is approximately linear. 

This circuit is not an attempt to achieve perfect linearity. It is a 
cheap method of improving the linearity obtainable in a commercial 
television receiver and has been proposed as a means of providing a 
sufficient degree of linearity without sacrificing the potential efficiency 
of the charging circuit. 

Tomlinson's Method of Linearization 

Tomlinson* has developed a circuit which, like Hawkins* circuit, 
employs an additional integrating circuit to improve the linearity. The 
basic circuit arrangement is shown in Fig. 87. 

In the case of a simple integrating circuit consisting of a resistance 
R x and a capacitance C 1 connected in series, to the input of which a 

steady potential is applied, the t •qh.t.+ 

current passing through the re- 
sistance and into the condenser 

necessarily falls exponentially as \ — P*Vt *ww— i — j-oOutput 

the condenser acquires a charge. 
Linearity of the rise of potential 
across the condenser can, more- 
over, only exist if the current 
through the resistance remains 
constant. One method of achiev- 
ing this result is to arrange that the steady potential across the 
integrating circuit is replaced by a potential which increases at the 
same rate as that at which the potential across the condenser increases. 
In this case, the potential across the resistance R Y remains constant and 
the current through it is likewise constant. Tomlinson has approached 
this result by inserting a further integrating circuit R 2 C 2 between the 
original circuit and the steady potential. The two integrating circuits 
must have the same time constant since the potential across C 2 is 
required to rise at the same rate as that across C v In this connection, 
two points emerge: 

(a) R 2 must approximately equal — - and C 2 must equal nC lt where 
n is >i. This is necessary so that the loading effect of R^i upon 

*T. B. Tomlinson: formerly Sobell Industries, Ltd., now Southampton Technical 
College. 



a- 



Si 

-WvW- 



fr Ti' .„. 



Fig. 87. Basic circuit of Tomlinson's 
method of linearization 



120 



TIME BASES 



C* 2 shall be reduced to a negligible amount since its impedance 
will then be high compared with that of C 2 . 
(b) The condenser C 1 cannot commence to charge until C 2 has 
acquired a potential, however small, to feed to the circuit /? J C 1 . 
This results in the commencement of the charging operation 
being non-linear, but the effect is not usually very serious. 

The two condensers must, of course, be discharged simultaneously 
by some device here represented by a switch across C 2 and a diode 
across R v When the switch is closed, C 2 is discharged and the junction 
of the two resistances is brought to the potential of the negative rail, 
while 6\ discharges through the diode. 

Since n is not easily made larger than 10 without drawing a large 
amount of current from the steady potential source while, at the same 
time, keeping the current into C 1 large enough to swamp any leakage 
current therein, it is advisable to adjust the value of R 2 for best results. 
Furthermore, if the condensers are not discharged to exactly the same 
potential, it becomes necessary to make a further small change in the 
value of one or other of the two time constants, and this is best achieved 
by adjustment of the resistance R 2 . If R x ^> R 2 , the potentials and 
currents are as shown in Fig. 87, and E represents the value of the H.T. 
potential applied to the circuit. 

e 9 = El 1 - 



R t i t = e(i - e~^c\ _ e v 



Now }-& 



*! + Jf2_ = JU* - f&\ 

dt R 1 C 1 R X C\ ' /' 

A e < e _J_ p / t ,r_i 1 -k 

_J e Rid j L_ , e K.d = _r Ij /e»c,_ e L/E.C, R,c t i\ 

dt R.C, R^K r 



and if R l C l = R 2 C 2f 

wV\ E 



= E (e *>c,\ _ E . t + k. 

\ / r x c; 

E L_ - '■ 

e, = E . t . e **c, _u k . e *>c, 






LINEARIZATION OF THE TRACE 121 

when t = o, e x = o and k = — E, so that 

1 -e RlCl )-ifr: '*' e RiCk 

(' \ E — -^— 

i _ e " el* \ and of .#.< Cl * 1 are plotted in 

Fig. 88, together with the curve of e x which is the difference between 
the two former curves. The improved linearity is evident from inspec- 
tion of the curves; it can be seen, however, that the commencement of 
the charging curve is non-linear. 




0f23 45 

Seconds 

Fig. 88. Characteristic curves pertaining to the circuit of Fig. 87 



Fig. 89 shows a practical arrangement of the circuit employing two 
thyratrons to discharge the condensers. The resistances R z and i? 4 , 
which should be about 500 ohms each, are intended to limit the anode 
current flowing through the thyratrons. The necessary bias and syn- 
chronizing potentials may be applied via i? 5 and i? 6 . The condenser C 3 
ensures that whichever valve strikes first will immediately initiate a 
discharge current in the other. 




GH.T.+ 



oOutput 



o/f.r- 

Bt'as 
o and 
Synch. 

Fig. 89. A practical form of 
Tornlinson's linearizing circuit 




vwvw- 



OH.T.+ 



o Output 



OH.T.- 

Bias 

■o and 
Synch. 



Fig. 90. Tornlinson's circuit employing 
a double thyratron 



122 



TIME BASES 



Fig. 90 shows how a double thyratron, if one existed, might be used 
to discharge the condensers, the simultaneous striking of the two 
portions of the valve resulting from the fact that the electrodes of 
each portion are immersed in the same gas. Thus, when current flows 
through one anode, the whole of the enclosed gas is ionized, both 
halves of the valve conduct and both condensers are discharged together. 

These circuits cannot give perfect linearity, but, neglecting the non- 
linear commencement, a very considerable improvement results from 
the application of the method, which may be compared with that of 
Fig. 112. 

Keen's Method of Linearization 

Keen* has developed a method of linearization which is similar to 
those of Hawkins and Tomlinson in that the result is achieved by 
means of an additional simple network. Several varieties of the circuit 
exist and these consist of capacitance and resistance or of inductance 



K 

l—^VWWWV— 1 

E Ci ± 
o 1 — 



Si 

-WWW 




(CL) 

e = 1 + pC t R 
E pCtR 

e = 1 + pCR % 
E 1 +pC (Ri+RJ 
or, alternatively, 

e = T^ 1 + hp 

E T-* + {k+i)p 

c. 



where T = C X R and k 



T = CR* and k 



R t 



ay 

for Fig. 91 (a) 
for Fig. 91 {/>) 

for Fig. 91 (a) and (b) 
for Fig. 91 (a), and 
for Fig. 91 (b) 



Fie. 91. 



Keen's linearizing circuits applicable to the basic CR 
integrating circuits 



and resistance networks. The correcting or linearizing networks which 
are applicable to the simple basic CR integrating circuit are shown in 
Fig. 91 and the circuits may be placed either before or after the basic 
circuit. The application of a unit function (see page 364) to either of the 
circuits of Fig. 91 produces the resultant wave-form shown in Fig. 92. 
It will be obvious that the rising response of the characteristic of 
Fig. 92 is of the proper form to provide a linearizing effect for a basic 
network preceded by one of the circuits of Fig. 91, provided the values 

*A. W. Keen: late of Sobell Industries, Ltd., now of E.M.I. Research, Ltd., 
Hayes, Middlesex. 



LINEARIZATION OF THE TRACE 



123 










-hi 



of T and k are correct since, as the potential across the condenser 
in the basic network rises, the input potential to the network also 
increases. 

The value of 2 1 for the correcting network should be the same as that 
for the basic integrating circuit 
and R x or C 2 may be varied to 
adjust the value of k. 

It is also obvious that either the 
correcting or the basic network 
may be placed at the beginning of 
the chain since, in general, both 
networks have high input and low 
output impedances. If there is 

any choice, it is better to place the basic integrating network at the 
input to the chain since it has a lower output impedance than the 
correcting network and is therefore less likely to be affected by the 
shunt impedance of the other network. 

There are five forms of the complete chain of networks, and these 
are shown in Fig. 93. In this figure (c) is derived from (a) by trans- 
forming the A connection of the three condensers into a Y connected 
system. The values of the condensers require to be altered. 



Fie. 92. Characteristic curves applicable 
to Keen's linearizing circuits 



I 



I 




Correcting circuit follows basic circuit, 
o — vwwvw — t — www 1— o 



C-AWWW 1 
-H-plH 

I— 



(C) 
Star-delta transformation of circuit (a.). 

n 1 1 O T I O 

CoV) (e) 

Correcting circuit precedes basic circuit. 

Fig. 93. Combinations of the basic CR integrating circuit with the two forms 
of Keen's linearizing circuit, Basic circuit is shown in heavy lines 



124 



TIME BASES 



The condensers forming part of the network must be discharged 
before the commencement of each new charging regime. In the case 
of Fig. 93 (b) and (e) the circuit arrangement required for discharging 
the condensers is more simple since only two condensers are employed. 




(cc) (b) 

Correcting circuit follows basic circuit. 

O — TfffWBU 1 — T TSSMP 




Star- de Ita transform at ion o f circuit (a.). 




(d) _ (e) 

Correcting circuit precedes basic network. 

i 



- — o 



(f) 
Modified method of drawing circuit (e) to show 
that a transformer may be employ ed, 

FlG. 94. Combinations of the basic integrating circuit with Keen's linearizing 
circuit. Basic circuit in heavy lines 

By replacing R by L and C by R a new series of networks containing 
only inductances and resistances is obtained. The five networks obtained 
by this transformation are shown in Fig. 94. In this figure (e) has been 
redrawn as (/) to draw attention to the fact that a transformer may be 
employed for integration. This provides a sixth form of the circuit. 

(c) LINEARIZATION BY MEANS OF A VARIABLE 
INDUCTANCE 

A number of workers have developed circuits for electromagnetic 
scanning in which the linearity is improved by the insertion of a gapped 



LINEARIZATION OF THE TRACE 



125 









iron-cored choke in the circuit. Owing to the partial saturation of the 
magnetic core, this choke has a variable value of inductance which is 
dependent upon the current flowing. The relationship between the 
current and the inductance is made linear over the working range by 
employing a small air-gap and the instantaneous value of the total 
inductance in the circuit is made to coincide, as nearly as possible, with 
that given by the equation; 



= L »(' - §} 



This equation states the conditions required to produce linearity, i.e. 
to ensure that the value of the inductance changes in such a way as to 
make the current rise or fall at a constant rate. In the above equation, 

L„ is the total initial inductance of the circuit (including the choke 
and any other inductance, e.g. scanning coils) as measured on an 
A.C bridge without D.C. polarization, 

L the instantaneous value of the total inductance, 

R the total resistance of the circuit, 

E the applied constant H.T. potential, and 

1 the instantaneous value of the current. 

Note that the value of R is assumed to remain constant although, in 
practice, an appreciable variation may occur because the anode im- 
pedance of the valve is not independent of the value of the anode 
current. 

The method described above is due to Fa rns worth,* but the equation 
which states the required relationship between the values was first 
calculated by Wilson, | and the circuit arrangements were further 
modified by Taylor} and Wood,§ who have developed other types of 
linearized blocking oscillators. 

Wilson's Method of Linearization 

Fig. 95 shows one form of the arrangement due to Wilson. The 
insertion of the gapped choke L 3 results in a decrease in the total 
inductance of the circuit as the current rises, so that the current con- 
tinues to rise at a constant rate. If the total inductance in the circuit 
is made to remain constant, the resulting current rises in an exponential 
manner. 

In this time base, the grid is biased at zero, or at a positive, potential 
during the forward stroke with the result that the anode impedance R a 
of the valve acquires a minimum value. During this time, therefore, 
the potential E is mainly divided between the deflection coils L t and L % 



* P. T. Farnsworth: U.S. Patent 2059683. 
X M. K. Taylor: British Patent 493142. 



f J. C. Wilson: British Patent 480754. 
§ H. Wood: British Patent 506856. 



126 



TIME BASES 



and the variable inductance L 3 . As the current rises, the value of the 
inductance of L 3 falls at a predetermined rate and allows the current 
to rise linearly. The flux produced by the deflecting coils is pro- 
portional to the current flowing therein and, as a result, the rate of 
deflection of the spot, which is proportional to the rate of increase of 
flux, is constant. The current is interrupted at the end of the stroke 
by a negative control signal. 

When employing this circuit at a repetition frequency of approxi- 
mately 10,000 cycles per second for the line time base deflection of a 



Deflection 
Coils 



- / 5iSW — TW- 
Li L 2 






■OH.T.+ 



Control 

Potential 



n 



-QHX- 



Fig. 95- 



Wilson's method of linearization by means of 
a variable inductance 



television receiver, the following values for the component parts may be 
employed: 

R a (anode impedance of the triode during the forward stroke) 

less than 1 ,000 ohms, 
R total (including R a ) about 1,200 ohms, 
L x and L 2 100 mH total (in situ), 
X, 3 varying from 60 mil with no D.C. to 20 mH with a current 

of 120 mA, and 
H.T. approximately 300 volts. 

Care must be taken to sec that the cathode ray beam is automatically 
"blacked out" if the control signals cease, as otherwise the screen of 
the cathode ray tube may be "burnt" by the stationary spot. In the 
original circuit, as developed by Farnsworth, the potential applied to 
the grid of V x is obtained by using the valve also as a squegging oscillator 
under the time control of a synchronizing signal whose amplitude can 
be considerably less than is necessary in the case of Fig. 95. 

Taylor's Method of Linearization 

The mechanism of one form of the linearized blocking oscillator may 
be understood from an examination of Fig. 96, which shows one of many 
circuits developed by Taylor. Just before the arrival of a synchronizing 
signal, the grid of V x is held negative relative to the cathode, by a charge 
accumulated on C x during the sweep, so that no anode current flows. 






LINEARIZATION OF THE TRACE 



127 



When the signal is applied via a third winding on the transformer, or 
by means of a resistance in series with the grid winding of the trans- 
former, the grid is driven rapidly positive and anode current commences 
to flow from the H.T. supply rail. At the beginning of the sweep, the 
anode potential across the valve is reduced to a low value but, as the 
current increases, the inductance of the choke falls, the anode potential 
rises and the current increases still further. The repetition frequency is 
controlled, in the absence of a synchronizing signal, by the values of 
R x and C v 

In Farnsworth's circuit, the coils L x and L 2 and the condenser C 2 
are connected across a resistance placed in series with the positive H.T. 
supply rail so that the coils, saturating choke and transformer primary 



OH.T.+ 




QH.T- 



Fig. 96. Taylor's method of linearization by means of 
a variable inductance 

are all in series with the anode of the valve. The choke has an in- 
ductance of about 150 henries when no steady current flows. Owing 
to the large flow of grid current, the impedance of the primary winding 
of T x is greatly reduced during the working stroke so that the greater 
part of the potential drop appears across L 3 . The inductance of the 
coils is made sufficiently low to permit the current to rise almost linearly 
when the potential rises in a linear manner. Farnsworth also employs 
a condenser and resistance connected in series between the anode and 
cathode of the valve to limit the rise of potential when anode current is 
cut off, and this procedure is generally advisable. 

In addition to the circuit shown in Fig. 96, Taylor has disclosed a 
number of other circuits in which the three inductive components 
(deflector coils, primary winding of the transformer T x and gapped 
choke) are not all connected in series as viewed from the valve. For 
example, the deflector coils, in series with a blocking condenser, may be 
connected directly between anode and cathode, the choke L z being 
connected in series with the primary winding of the transformer.* The 
operation of the circuit then proceeds as follows: The gapped choke 

* See Fig. 2 of British Patent 493142. 



128 



TIME BASES 






controls the rate of change of the primary current in the transformer 
and therefore governs the rate of rise of grid potential. This, in turn, 
determines the rate of change of anode current which is divided between 
the deflector coils and the transformer. In this way, partly by design 
and partly by experiment, it is possible to arrange that the anode current 
variations shall provide a substantially linear change of current through 
the deflector coils. Many other arrangements are possible, and these 
are shown in the patent specification to which reference should be made 
if further information is required. Benjamin's* paper should also be 
consulted. 

(d) LINEARIZATION BY MEANS OF FEED-BACK 
ARRANGEMENTS 

The linearity of a saw-tooth time base may be greatly improved by 
the application of feed-back. f However, it must be stressed that this 
improvement may be difficult to realize in practice unless special 
precautions are taken to avoid distortion. It has already been mentioned 
that a high-impedance constant-current charging circuit will not pro- 
duce a linear rise of potential across a condenser which is shunted by 
a resistance, even when the shunt resistance is connected through a 
blocking condenser (see page 18). The reason is that the blocking 
condenser becomes charged to the mean potential of the saw-tooth and 
acts as a second source of supply to the time base condenser. Thus, 
current flows from the blocking condenser to the time base condenser 
when the latter is charged to less than the mean potential. When the 
time base condenser becomes charged to a higher potential than the 
blocking condenser, current flows in the reverse direction. The net 
result is that the charging curve becomes exponential. It is rather 
inconvenient to connect the time base condenser directly to the deflector 
plates of a cathode ray tube because the trace is usually required to start 
at the left-hand edge of the screen. For this reason, zero potential across 
the time base condenser will not correspond with zero potential between 
the plates, and it becomes necessary to introduce a steady shift potential. 
In order to permit the use of a shunt resistance without interfering with 
the linearity, the time base condenser may be isolated from the deflector 
plate circuit by means of a buffer stage, which, in order to obtain a low 
value of output impedance and so to reduce distortion to a minimum, 
should preferably be of the cathode follower type. When this isolation is 
provided, it becomes practicable to improve the linearity to almost any 
desired extent provided the insulation of the main time base condenser 
and the associated circuit is of the order of 1,000 megohms. 

* R. Benjamin: "Blocking Oscillators," Journ. I.E.E., 1946, 93, PartlllA, p. 1159. 
t H. S. Black: "Stabilized Feed-back Amplifiers," B.S.TJ,, 1934, 13, p. 1. 



LINEARIZATION OF THE TRACE 



129 



The Linearization of Electrostatic Deflection Potentials 

by Means of Negative Feed-back 
The separation of the time base condenser from the stray capacitance 
of the deflector plate circuit, by means of a cathode follower, has the 
result that there are two capacitances to be charged (or discharged) 
during the fly-back period. One of these capacitances is that of the 
main charging condenser, while the other is the capacitance appearing 
across the cathode load, i.e. that due to the cathode itself and to the 
deflector plates of the cathode ray tube. The circuit should therefore 
be arranged so that a single switching device will bring about a rapid 
charge of both these capacitances. A device of this nature will auto- 
matically result if the fly-back is arranged to drive the grid of the 
cathode follower in a positive direction. 
Norton and Groves* Method of Linearization 

There are two feed-back methods for obtaining a constant current, 
one of which operates by correcting for variations in the current and 
the other by introducing a compensating potential in the charging 
circuit to counterbalance the potential across the condenser. A simple 
example of the first method is the use of an unshunted cathode bias 
resistance in an otherwise 
ordinary pentode charging cir- 
cuit. That it is possible to im- 
prove the linearity in this way 
was first pointed out by Norton 
and by Groves.* The effective 
anode to earth impedance of a 
valve with a resistance R 1 'm the 
cathode circuit is R a -{-R 1 (1 -\-[i) 
where R a is the normal anode 
impedance of the valve and jx 
is the amplification factor. 

The value of the effective anode to earth impedance may be re- 
garded as a measure of the constancy of the anode current for a given 
change of anode potential. In order to permit a high value of cathode 
resistance /?j to be used without reducing the anode current, the grid 
of the valve should be connected to a tapping on the H.T. supply, as 
shown in Fig. 97, in which a switch is shown for discharging the con- 
denser. It is important that the anode current be kept reasonably high 
to minimize the effect of any leakage current. 

In order to obtain an idea of the magnitude of the effect of a feed- 
back resistance in the cathode circuit of a pentode, consider a valve 

* F. R. Norton: British Patent 467958; W. R. Groves: Letter to the Wireless World, 
1938, 43, p. 278. 



a Out 




0£ } * 



OH.Tr 



Fig. 97. 



Basic charging circuit employing 
negative feed-back 



no 



TIME BASES 



having an anode impedance of 2 megohms and an amplification factor 
of 2,000. Then, if R t is made 100,000 ohms, for example, the effective 
impedance as given by the above equation becomes 202 megohms 
instead of 2 megohms. Note that the circuit of Fig. 97 provides negative 
current feed-back and that this type of feed-back results in a high value 
of output impedance at the anode. Because the output impedance is 
high, the circuit will have its linearity reduced if leakage is present 
across either the condenser or the valve. 

The relation between the effective output impedance and the linearity 
of the potential wave-form across the condenser may be appreciated by 
calculating the supply potential required to pass the same current 
through an ohmic resistance equal to the effective impedance. If the 
pentode in the above example takes an anode current l a equal to 2 milli- 
amperes, then the current will remain constant to the same degree as if 
there were an ohmic resistance of 202 megohms and an H.T. potential of 
202 x 10 6 x 2 x io -3 =404,000 volts. Thus, the equivalent H.T. potential 
is ^aORa+[i+E*]#i) or approximately v.I a R v 

Without the feed-back resistance the equivalent potential is I a R a> 
which is only 4,000, and it is obvious that the current will vary quite 
appreciably as the time base condenser becomes charged. For instance, 
the current will fall by 5 per cent when the condenser is charged to 
5 per cent of 4,000 volts, i.e. when it is charged to 200 volts. For the 
same potential sweep the negative feed-back arrangement maintains the 
current constant within 005 per cent, i.e. 200 volts in 400,000. From 
these examples it will be seen that the product of the current and the 
effective impedance gives a direct indication of the value of a constant 
current device. 

In the method described above the negative feed-back resistance R x 
supplies a controlling potential to the cathode of the valve and this 
regulates the current into the condenser. The control potential changes 
only because the current changes slightly. 

Busse, van der Mark and Venis* have patented a time base in which 
the linearity is improved by raising the value of the charging potential 
as the condenser charge proceeds. 

The condenser is charged through a resistance and a two-valve 
resistance-capacitance coupled amplifier is connected across the con- 
denser. The resistance in the anode circuit of the second valve is in the 
form of a potentiometer and the condenser charging resistance is con- 
nected to the slider. Thus, as the condenser potential increases, the 
potential at the anode of the last valve increases so that the charging 
potential at the slider of the potentiometer also increases. By adjusting 
the position of the slider, the potential employed to charge the condenser 

* E. Busse, J. Van der Mark and A. Venis: British Patent 452965. 






LINEARIZATION OF THE TRACE 131 

can be made to increase at the same rate as that across the condenser so 
that the charging current may be made to approach a constant value. 

Newsam's Method of Linearization 

The principle upon which Newsam's* method of linearization 
operates is that of the cathode follower and it may be understood by 
considering the time base circuit shown in Fig. 98. It will be seen that 
two H.T. supplies are involved, but it will be shown later how this 
inconvenience may be avoided. 

The operation of the circuit is dependent upon the fact that the 
cathode potential follows the varying grid potential as the condenser 




oHZ- 

Fig. 98. A cathode- follower constant- current time base circuit 

C l charges; the circuit is a cathode follower. The upper plate of the 
condenser C ± acquires a positive charge with respect to earth. Since 
the valve acts as a cathode follower, the potential difference between 
the grid and the cathode is negligibly small compared with that of the 
battery E 2 . Consider the closed circuit C u R lt E 2 , R 2 . In a closed 
circuit the sum of all the potentials is zero : therefore, commencing at 
the negative rail and working in an anti-clockwise direction, we have 

&R2 "H *jbj Ari =s ^ci» 
where E R2 is the potential across R m and so on. 

Now, if the valve were a perfect cathode follower, i.e. if there were 
no difference in potential between the grid and the cathode, the potential 
across R z would be identical with the potential between the grid and 
the negative rail, i.e. with that across C x . Cancelling out E R2 and E C1 
in the above equation gives 



**Bi ~~ **JC1 



Eg 



;. the current in R x = ~ which is constant. 
R 1 

It is thus evident that the more perfect is the behaviour of the valve 

* B. Newsam; British Patent 493843. H. L. Krauss: "Graphical Solutions for 
Cathode Followers," Electronics, 1947, p. 116. 



132 



TIME BASES 



and its associated circuit as a cathode follower, the more perfect will 
the linearity become. This argument is, of course, subject to the 
limitation that the charging rate will become non-linear when the 
potential of the grid approaches too closely to that of the anode. 

It can be shown that for circuits of this type the linearity is the 
same as if the condenser C x were being charged through an ohmic 
resistance from a high- potential source of supply. In the particular 
case of a triode valve circuit with R x made equal to R a , the equivalent 
H.T. supply potential is 

— — E 2 -\-\E x approximately. 

2 

It appears that if a pentode having an amplification factor of 2,ooo 
were used, the equivalent potential would be in the region of 1,000 E 2 . 

o£ 2 + 



& 



7 



c 2 



o 



AY 



c 2 



oOut 



OKZ- 



Fig. 99. An improved form of the circuit of Fig. 98 

In order to employ a pentode in the circuit of Fig. 98 it would only be 
necessary to connect the screen to a suitable tapping on the battery E 2 . 

Several variations of this circuit are possible and one may so arrange 
matters that the necessity for two distinct sources of supply is avoided. 
In a circuit developed by Newsam, a substitute for the battery E 2 is 
employed. The basis of this scheme is illustrated by the simplified 
diagram Fig. 99, from which it will be seen that the place of E 2 in 
Fig. 98 is now occupied by a condenser C 2 . The capacitance of C 2 is 
made sufficiently large that the potential across it remains steady and 
it is kept charged from the main H.T. supply through R A . It is to be 
noted that the cathode load impedance now consists of R 2 in parallel 
with R 3 (in the A.C. sense). 

Instead of taking the output potential across the condenser, the 
deflector plates may be connected across R 2 or j? 3 so that the valve not 
only behaves as a cathode follower linearizing device but it also func- 
tions as a cathode follower buffer stage between the time base condenser 
C x and the deflector plates. It is then possible to use any desired 









LINEARIZATION OF THE TRACE 



133 



arrangement of the X-shift without fear that the shunt resistance will 
destroy the linearity. 

It is a disadvantage of the particular arrangement shown in Fig. 99 
that the grid is driven negative during the fly-back period, because 
there will usually be sufficient stray capacitance associated with the 
output circuit to delay the fall of cathode potential. In consequence 
the valve becomes momentarily cut off and a relatively slow fly-back 
is obtained. For this reason it is preferable to arrange the circuit in 
such a way that the grid is driven positive during the fly-back period. 
This may be carried out by disconnecting R z from £ t + and connecting 
it to an additional source of potential E t — , as shown in Fig. 100. If the 



0%+ 



• f-^vvvvv^- 



Vt 



(T 






-oOut 



■0H.T.+ 



-o£o~ 



Fig. 



100. Modified form of the circuit of Fig. 99 arranged to cause 
the grid of Vi to be driven positive during the fly-back 



rate of change is to remain unaltered, the potential E t must be equal 
to that of E x . The discharge device, represented by the switch, must 
then have its cathode connected to the upper end of C x and its anode 
to the lower end. In order to prevent ripple from the heater supply 
from appearing at the grid of the triode V Xl it may prove necessary to 
employ a steady current for heating the cathode of the discharge device 
or to adopt the artifice described near the top of page 41 . 

A push-pull version of Fig. 99 is given in Fig. roi. This circuit is 
really a combination of the feed-back circuit with the well-known 
phase-reversing circuit, as shown in Fig. 137, page 197. The anode load 
resistance R 4 is made equal to the cathode load, i.e. 

_ R 2 R 3 

It will be seen that the grid is connected to the condenser C x via a 
blocking condenser C 3 so that the grid bias may be independently 
controlled by the tapping on R 2 . The presence of R A somewhat impairs 
the efficiency of the valve, considered as a cathode follower and, of 
course, this circuit also suffers from the fact that the grid is made 
negative during the fly-back period. Nevertheless, it appears that the 



134 



TIME BASES 






principle of employing a cathode follower to provide linear charging 
and simultaneously to function as a buffer valve and phase-reverser, is 
likely to be very useful when accurate linear time bases are required. 

It is well to consider the similarities and the differences between the 
foregoing circuits which have been developed to provide increased 



0£ z + 




OffX- 

Fig. 101. A cathode-follower constant-current time base 
circuit providing push-pull output 

linearity by taking advantage of the benefits to be gained by the 
application of the feed-back principle. 

Figs. 98 to roo inclusive employ negative voltage feed-back: close 
inspection of the circuits will show that the condenser C t is, in fact, in 
each case, connected between the anode and the grid of the valve. To 
clarify this point, the circuit of Fig. 98 has been redrawn as Fig. 102 and 




xi^-tft? 



;*2 



Fig. 102. 

Fig. 98 redrawn. A cathode-follower 

constant-current time base circuit 



Fig. 103. 

The first step in the modification 

of Fig. 98 



Figs. 103, 104 and 105 show progressive changes in the circuit. The* 
result shown in Fig. 105 should be compared with Fig. 122 (Miller- 
capacitance Time Base) on page 159. 

The (anode) output impedance of a circuit employing negative 
voltage feed-back is low. In order to prove this point, assume that a 



LINEARIZATION OF THE TRACE 



*35 



small potential change Se 2 is suddenly applied to the anode of Fig. 105, 
then the potential change at the grid will be 

Be. = Be 2 MCi+CJ*. , 

where C s is the stray grid-cathode capacitance. The change of current 
in the valve resulting from the above change of grid potential Is g Se ff , 
where g is the slope in amperes per volt and this current will be drawn 



1 






\R 2 



\Ri 



-0S2-1- 



Ci 



Fig. 104. 

The second step in the modification 

of Fig. 98 



*5 

Zz- 

°*i- 

Fig. 105. The final step in the modification 

of Fig. 98 (compare this circuit with that of 

Fig, 122, page 159) 



from the source of potential Se 2 . The impedance looking into the anode 
is therefore: 






1 +/»[C 1 + CJi? J 1 



P [C, + CJ R L g 

For a normal receiving valve, the value of ijg is approximately 
150 ohms and, as a result, the output impedance is quite low. Even if 
p\C x + C s ] i?! is small compared with 1, the impedance will not have a 
high value. It will be seen from this argument that a low anode output 
impedance results from negative voltage feed-back. Compare this 
remark with that on page 130, which states that the anode output 
impedance of a circuit employing negative current feed-back is high, but 
note that the cathode output impedance is low. 

It should be appreciated that the degree of linearity of all these cir- 
cuits embodying Newsam's method of linearization depends upon the 
extent to which the value of the gain of the cathode follower valve can 
be made to approach unity. On the other hand, the circuit has the 
great advantage that it has no tendency to instability. 

The Bootstrap Circuit 

A further modification of Newsam's circuit, in which the resistance 
/? 3 of Fig. 99 is replaced by a diode, has been developed by Williams.* 

* F. C. Williams: "An Introduction to Circuit Techniques for Radiolocation," 
jfoum. I.E.B., 1946, 93, Part I HA, p. 289. 



136 



TIME BASES 



The circuit which is shown in Fig. 106 has been called a bootstrap 
circuit for the obvious reason that as the potential at the left-hand end 
of R x rises, that at the right-hand end rises by an equal amount, as 
already explained on page 131. For this reason, the arrangement is said 
to "lift itself by its own bootstrap."* The name is, however, equally 
applicable to the circuits of Figs. 98 to 105 inclusive. 

The valve V ly to the grid of which a square wave-form is applied, 
functions as a discharge device for the time base condenser C x . In 
order to discharge the condenser to the greatest possible extent, the 
grid of V l should be driven a few volts positive by the square wave- form 



0+200 v 



Control 
Potential 




o OV 



0-120V 



Fig. 106. The bootstrap circuit 



control potential. Furthermore, the amplitude of this square wave-form 
should be sufficient to cut the anode current in V x completely off during 
the whole of the time while the condenser is charging and should not 
be less than, say, 25 volts, i.e. from +5 to —20 volts. The rate of rise 

J? F 1 

of potential across C x at the commencement of its charge is '" 6 , 

R i C 1 

where E a is the potential of the positive supply rail and E b is that of 
the anode of K x at the instant before the application of the negative 
switching pulse. Before V x is cut off, the diode V 2 is conducting but, 
when Cj commences to charge, the potential of the cathode of F 3 rises 
(F 3 acting as a cathode follower) carrying with it the cathode of V z 
which thereupon ceases to conduct, the condenser C 2 having been 
charged via F 3 and R, z to the potential of the supply rail. This con- 
denser C 2 then behaves as a battery and supplies the necessary potential 
to charge C x via R lt the output being taken from the cathode of F 3 . 
Further methods of linearizing capacitive time bases by feed-back 
methods are described on page 140 and in Chapter IX. 

•The expression "bootstrap" is more familiar in the U.S.A. than in this country. 






LINEARIZATION OF THE TRACE 



i37 



The Linearization of Electromagnetic Deflection Potentials 

by Means of Negative Feed-back 
Circuits for the application of feed-back linearizing methods to 
electromagnetic time bases have been developed by a number of workers. 
One form of circuit is due to Blumlein* and, in part, also to Bowman- 
Manifold, Percival and White. Newsam,f Bedford,} Bull,§ and Faudell|| 
have also developed circuits for this purpose. 

Blurnlein's Method of Linearization 

Blurnlein's circuit is shown in Fig. 107, in which the condensers C 3 
and C 9 are charged in series via the resistance R z when V x is cut off 
and are discharged rapidly when a positive synchronizing pulse is 

— OHT.+ 




-o/y.?:- 



Fig. 107. Blurnlein's method of linearizing the current in the deflector coils of a 

capacitive time base 

applied to the grid of V x . The time constant C A R- a is made large so 
that the potential variations at the grid of V z are substantially equal to 
those at the anode of V v Although the current through R 3 charges 
C 3 and C 9 in series, it does not produce a large change of potential at 
the grid of V 2 , because the feed-back resistance R d passes an opposing 
current into C 9 . As soon as the grid of V 2 commences to go positive 
as C 3 is charged, the anode potential of V 2 falls and the junction of C 3 
and C 9 becomes rapidly negative. Since the potential at the grid of F 2 
is nearly constant, the potential across R 3 is nearly constant and the 
arrangement forms a constant-current charging device due to the 
application of negative feed-back. The condensers C 3 and C 9 are 
therefore charged in a nearly linear manner. It must be remembered 
that, by definition, negative feed-back is a circulation of energy or 

* A. D. Blumlein: British Patent 4791 *3 ; M. Bowman -Manifold and W. S. Percival: 
British Patent 424221 ; E. L. C. White: British Patent 318378. 
t B. Newsam : British Patent 493843- 
X A. V. Bedford: British Patent 401990. 
§E. W. Bull: British Patent 522637. 
I1C. L. Faudell: British Patent 51 1847 (see page 140). 



i 3 8 



TIME BASES 



(a>) 




potential in a closed circuit such that the energy or potential at the 
input is greatly reduced but remains operating in the same sense 
as without the feed-back. Since the current through R 3 is almost 
constant, it follows that the current through R 9 flows upwards and 
is approximately constant during the trace, while it flows down- 
wards and has a larger amplitude during the fly-back period. The 
potential developed across R 1Q during the trace consists of a constant 

part equal to iR$ plus a linear change equal to n n • The wave-form 

of the potential appearing across R l0 is therefore of the type shown in 

Fig. 108 (b) which is the potential wave-form necessary to produce a 

saw-tooth current wave-form 
in an inductance which pos- 
sesses series resistance. In 
the present instance, the de- 
flector coils are energized via 
the transformer while the 
feed-back circuit is arranged 
to compensate the complex 
network constituted by the 
leakage and primary induc- 
tances and resistances of the 
transformer plus the induc- 
tance and resistance of the 
deflector coils including the 

connecting cables. This equivalent network is shown in Fig. 109, 

together with the complementary network. 

R T represents the primary resistance of the transformer, 

L T the primary inductance, 

R c the resistance of the secondary winding of the transformer plus 
that of the cable and deflecting coils, and 

L c the leakage inductance of the secondary winding of the trans- 
former plus the inductance of the cable and deflecting coils. 

It is to be noted that, in the compensating network, the condenser 
C 3 is shown connected in parallel with C 9 , although the dotted con- 
nection is absent in Fig. 107. Since the grid potential of V 2 is held 
almost constant by the feed-back process, the condensers C 3 and C 9 
are effectively in parallel as regards the impedance in series with the 
resistance i? 9 . 

In order to determine the values for the component parts of the 
compensating network, the following equations are employed : 



Fig. 108. Potential wave- form required to pro- 
duce a saw-tooth current wave-form in (a) a 
pure inductance, {b) an inductance possessing 
resistance 



LINEARIZATION OF THE TRACE 



139 



(C 3 +C 9 ) ic 9 = p" 



R 9 LcNi* 
'~L T NJ' 



L 9 
*10 



where A/\ is the number of turns on the primary winding and 
AT 2 is the number of turns on the secondary winding. 

■AAAAA— f 1|- 



R 



■w 




Comp/ementary Network 



Network representing 
Transformer and 
Def/ecting Coifs 

Fig. 109. Networks associated with the time base of Fig. 107 

In order to solve all the above equations, the values of C 3 and C 9 
must be determined and these values must be such that the capacitance 

C£s_ .^ together with the values f Rs an d t he H.T. potential, 

O ■» ~x~ (-* j) 

suitable for the required trace duration. C 3 and C 9 should preferably 
be made equal in value. 

In general, the compensation obtained will be imperfect, but adjust- 
ment of R 9 will control the fly-back wave-form while the linearity of 
the trace may be controlled by adjustment of /? 10 . Note that R 10 and 
C 10 are only required for low repetition frequencies such as are met 
with in television frame-scanning time bases. 

The circuit of Fig. 107 shows a cathode bias circuit C Z R % designed 
to hold the anode current cut-off except during the synchronizing 
impulses. If the circuit is to be used with a blocking oscillator such 
as V x of the Holmes, Carlson and Tolson time base of Fig. 67 shown 
on page 94, the grids of the two valves should be connected directly 
together and C v C 2 , R t and R 2 of Fig. 107 should be omitted because 
the grid of the blocking oscillator is held negative for the duration of 
the trace. 
FaudelTs Method of Linearization 

Faudell* has developed two versions of a time base employing two 
*C. L. Faudell: British Patent 511847. 



140 



TIME BASES 



valves, in one of which anode-grid coupling is employed for each valve 
and, in the other, anode-grid coupling is used for one of the couplings 
and cathode coupling for the other. The circuit employing only anode- 
grid coupling is shown in Fig. no. 



OH.T+ 



Output 




Fig. 1 10. Faudell's method of linearization employing only anode-grid coupling 

In the preferred form, the circuit which behaves as a feed-back 
linearized multivibrator, when V± is non-conducting and V 2 is con- 
ducting, the condenser C 2 becomes charged with its left-hand plate 
positive with respect to the other and, when the valves reverse their 
roles, the charge in C 2 is reversed. The condenser C 2 applies negative 
feed-back from the anode of each valve to its own grid and so linearizes 
the output current (see page 158). The circuit thus contains two 
feed-back circuits of which one is formed by the anode-grid couplings 
and is employed to cause the circuit to oscillate while the other is 
formed by the presence of the condenser C 2 to provide a capacitive 
coupling from each anode to its own grid for linearizing purposes. 



OH.T.+ 




Control 
Potential 



Output 
o 



Fig. 111. Faudell's method employing one anode-grid and one 
cathode coupling 

In the circuit of Fig. in, no linearizing feed-back circuit is em- 
ployed. If the control potential input terminals are short-circuited, 
the circuit will oscillate in much the same manner as does Potter's 



LINEARIZATION OF THE TRACE 



141 



time base. Synchronizing potentials may be applied to the control 
potential terminals or to the anode of the valve V 2 . 

The primary winding of the output transformer T x should have a 
value of resistance which is suitable for correctly biasing the grid of the 

valve V 2 . 

As explained on page 76, the grid leaks in either of these circuits 
may be connected, e.g. to the positive H.T. supply rail in order to reduce 
the effects of small changes of potential, or of the valve characteristics, 
upon the stability of the oscillators. 

The Linearization of Electrostatic Deflection Potentials 
by Means of Positive Feed-back 
Beale and Stansfield's Integrating Amplifier 

Another method of forcing a constant current into a condenser 
employs positive feed-back. The potential across the condenser is used 
to control a valve which attempts to inject an exactly equal potential 
in series with the source of supply so as to keep the potential drop 



In 



Ri 

O -. o-l WWW 

First 



rfi 



Stage 



C x 



I Second 
Stage 



o-~ 



i*, c 



Out 



Fig. 112. Block circuit diagram of Beale and Stansfield's integrating amplifier 

across the charging resistance independent of the accumulated charge 
on the condenser. 

This principle was first used in an integrating amplifier developed 
by Beale and Stansfiefd.* In their circuit a two-valve amplifier is 
arranged with an elementary integrating circuit between two amplifying 
stages, as shown in Fig. 112. A small fraction of the amplified condenser 
potential is taken from the output potentiometer and is fed back in 
series with the input potential. If the feed-back resistance is carefully 
adjusted, almost perfect integration may be obtained over a wide range 
of frequencies in the absence of valve distortion. Theoretically, one 
would expect this adjustment to coincide with that at which the circuit 
becomes unstable but, in practice, the amplifying stages and the feed- 
back channel always contain resistance-capacitance couplings which 
reduce the gain at very low frequencies and so prevent self-oscillation 
of the circuit in normal use. 

In practice, a more accurate result may be obtained by using a 
device which is not capable of neutralizing entirely the potential across 
the condenser. 

* E. S. L. Beale and R. Stansfield: British Patent 453887. 



142 TIME BASES 

METHODS OF MEASURING THE LINEARITY AND 
VELOCITY OF A TIME BASE TRACE 

It is essential for many measuring purposes to know the velocity of a 
time base trace. Furthermore, without a knowledge of the velocity of 
the trace, no information can be derived from an oscillogram other than 
the amplitude and wave-form of the signal under examination. When 
considering the linearity of a time base trace as distinct from a time base 
wave-form, it is essential to remember that the trace will contain errors 
which are not present in the wave-form. These errors are due either 
to imperfections in the cathode ray tube or to the way in which the tube 
is viewed: 

(a) Unless the angle of deflection is inconveniently small, it is very 
difficult to obtain strict proportionality between the angle of 
deflection and the movement along the curved surface of the tube. 

(b) The angle of deflection may not be accurately proportional to 
the deflecting potential, e.g. in the case of gas-focused tubes, 
origin distortion will usually introduce this type of error.* 

(c) The thickness of the glass wall introduces an error unless the 
screen is viewed normal to its surface. 

These errors may differ for the X and Y directions of deflection. 
In the case of (a), one cause of the difference is that the length of the 
deflected portion of the beam is not the same for the X and Y directions 
since the X and Y plates are normally at different distances from the 
screen. In the case of (b), the error, if it exists, is likely to differ for 
the two directions of deflection. In the case of (c), the error is de- 
pendent on the position of the eye of the observer with respect to each 
individual part of the image. 

The combined effect of all the above errors is likely to be of the 
order of I per cent of the total length of the trace. 

A time base is used for measuring time intervals and, if it is intended 
to make these measurements on the screen of a cathode ray tube, the 
measurement of linearity will normally be made in the same way so 
that any distortion in the tube will be combined with the error of the 
time base itself. When the time base is used for time measurements by 
other methods, i.e. without a cathode ray tube, it is usually required to 

* M. von Ardenne: "Eine neue Methode zur Beseitigung der Verzerrungen durch 
Raumladung in Braunschen Rohren" ("A new Method of avoiding Space-charge 
Distortion in Cathode Ray Tubes"), Zeitschrift fiir Technische Physik, 1933, 14, p. 461. 

D. Robertson: "The Examination and Recording of the Human Electrocardiogram 
by means of the Cathode Ray Oscillograph," Journ. I.E.E., 1937, 81, p. 497; "Some 
Observations on the Adaptation of the Cathode Ray Oscillograph to the Recording of 
Bio-electrical Phenomena, with Special Reference to the Electrocardiogram," Proc. 
Royal Society of Medicine (Section of Physical Medicine), 1936, 29, p. 23. 



LINEARIZATION OF THE TRACE 



H3 






know the overall accuracy of the system, and this can be measured with- 
out analysing the errors. If the overall error in such a system is very 
small, it becomes difficult to determine what fraction of the error is 
really due to the time base itself, this error being often much smaller 
than the other errors in the system and, of course, the advantage to be 
gained by this knowledge becomes less valuable. In systems of this 
type, the saw-tooth wave-form is usually applied to a piece of apparatus 
which provides a signal when the time base output potential reaches a 
known value. This is precisely what is required in order to measure 
the linearity of the time base. However, if a better signal-generating 
device is invented it is immediately used, not for measuring linearity, 
but as part of the equipment itself and the problem of analysis remains. 

The Use of a Calibrating Oscillator for Measuring the Linearity 
of the Trace 

In order to measure the degree of linearity, one may employ an oscilla- 
tor whose output is suitably distorted and passed through a differentiat- 
ing circuit before being applied to the Y plates of the cathode ray tube, 
the time deflection potential being applied to the X plates. It is usual to 
employ a diode to remove alternate pulses since adjacent pulses have 
opposite sign and, hence, only tend to confuse the image. The differen- 
tiated output potential will produce narrow pulses, or spikes, on the 
time base trace and the time interval between them will be constant. If, 
therefore, the variation of distance between the pulses can be measured, 
the variation of trace yelority, and hence the degree of linearity, can 
be measured. The velocity of the trace is equal to the pitch of the 
calibrating marks multiplied by the frequency of the calibrator. The 
velocity may be as low as 1 inch per second or as high as several miles 
per second. The easiest method of measuring the distance between 
the pulses is to place a transparent measuring rule in front of the 
cathode ray tube. This rule should preferably consist of a narrow strip 
of transparent flexible material engraved in centimetres and millimetres. 
In order to avoid parallax, the rule should be placed with its engraved 
surface next to the screen of the cathode ray tube, the rule being bent 
to conform to the shape of the end of the tube. 

For certain purposes, the calibrating pulses or "pips" may perma- 
nently remain on the cathode ray tube screen while the apparatus is in 
operation in order to provide a "yard-stick" and, in this case, they may 
appear superimposed upon the wave-form under examination or they 
may be made to appear upon a separate parallel trace by employing a 
double-beam cathode ray tube. The use of a single trace, which is 
switched by means of a square wave generated by an electronic switch, 
to a different vertical level and fed with the signal and calibrator pips 



144 



TIME BASES 



during alternate traces is deprecated since the wave- form under exami- 
nation and the calibrator pips, as seen on the cathode ray tube, do not 
occur at the same time. The method is, therefore, capable of leading 
to unsuspected errors. 

In order to ensure that the pulses appearing on successive traces shall 
be superimposed upon one another it is necessary either that: 

(«) the time base shall be synchronized by the oscillator, or 
(b) the oscillator shall be quenched after each trace and restarted at a 
fixed time before the next trace. 

If the time base is synchronized by the oscillator, the degree of syn- 
chronization should be sufficiently small to avoid serious reduction of 
the amplitude of the trace. This is of importance in some, but not in all, 
time bases. This method is not suitable where the time base repetition 
frequency is fixed, although one may obtain any desired repetition fre- 
quency by setting the oscillator frequency to a harmonic of the time base 
frequency. The synchronizing control is then carefully adjusted until 
the correct number of calibrator pips is obtained on the trace. The 
correct number of calibrator pips is equal to the ratio of the oscillator 
and time base frequencies. 

If the repetition frequency is fixed and there is a sufficient delay 
between successive traces, it is advantageous to quench the oscil- 
lator and to restart it at, or just before, the commencement of the 
new trace. It is essential that the period during which the oscillator 
is quenched shall be sufficient to permit the oscillations to be reduced 
to an amplitude which is so small as not to affect the instant at which 
the oscillator recommences when the quenching signal is removed. 
If the time during which the oscillator is quiescent is too short to 
allow the amplitude to fall sufficiently, any variation in the duration 
of the quenching period will result in inaccurate superposition of the 
pulses. The quenching signal usually takes the form of a square wave 
and is obtained from a multivibrator or a flip-flop or by amplifying 
and squaring the actual time base wave-form. Unfortunately, this 
latter method necessitates the commencement of the calibrating oscilla- 
tions after the beginning of the time base trace so that the accuracy 
of the early part of the trace cannot be checked. 

One very important point in connection with the removal of the 
quenching signal is the effect of the slope of the wave-front upon the 
output frequency during the first few cycles. If the signal which 
restarts the oscillations does not attain its final level within a time which 
is small compared with the period of oscillation, the frequency may vary 
during the first few cycles. One explanation for this is that the grid- 



LINEARIZATION OF THE TRACE 



145 



Fig. 113. Ideal sloping linear 
wave-front 



cathode capacitance of the valve depends upon the instantaneous value 
of the grid potential. If this capacitance appears across the whole or 
part of the tuned circuit, the frequency will obviously vary during the 
transition. Another contributing factor may be stated thus: if the con- 
trolling wave-form is considered as an 
ideal sloping linear wave-front, as in 
Fig, 113, applied to an oscillating 
circuit, the latter will receive two 
kicks, each of which is capable of 
producing shock excitation — one at 
the beginning of the wave-front and Time 

one at the end. The second kick is 
unlikely to occur at the precise instant 
which is necessary to maintain the 

oscillation in its original phase. In practice, the wave-front will not 
be linear and energy is transferred more gradually, the result being 
even more complex. 

The effect of the frequency variation during the first few cycles is 
most easily avoided by arranging that the oscillation is started in time 
for the cycles of irregular frequency to be completed before the time 
base trace commences, where this is possible. For this purpose, the 
oscillator should be made to function for at least five cycles before the 
commencement of the trace. Such an arrangement is only possible 
when there is a considerable waiting period between the traces, and it 
may be carried into effect by the methods described in Appendices VI, VII 
and VIII. If the interval is too short to permit this course of action to be 
taken, the linearity of the early part of the trace cannot accurately 
be checked unless the oscillator can be switched on more rapidly by an 
increase in the steepness of the switching signal The attainment of 
this end is also assisted by reducing the value of Q for the tuned circuit 
and, at the same time, the quenching action is improved because the 
oscillations are thereby made to decay more rapidly. The Q value may 
be reduced by connecting a resistance in shunt with the tuning coil. 
This resistance should be made sufficiently low to cause rapid decay 
and build-up of the oscillations but not low enough to reduce the 
amplitude of the oscillations unduly. An improved method employing 
a beat frequency oscillator is described on page 155. 

The circuit of a suitable oscillator is shown in Fig. 114, from which 
it will be seen that V x is a pentode valve which oscillates because of 
the presence of positive feed-back between the anode and the screen- 
grid, the control grid being employed for quenching the oscillations 
between successive time base traces. In order to obtain a clearly defined 
signal, the oscillations are squared by applying them to the grid of V tt 



146 



TIME BASES 



whose grid base potential is small compared with the potential applied. 
The output of V 2 is then differentiated by means of the short time- 
constant network C 5 i? 6 and the resultant peaky signal is applied to 
a Y plate of the cathode ray tube. The input circuit to the valve 
V 2 may, if desired, be modified in accordance with the information 
contained in Appendix VI. 

The time interval between the calibration pulses on the time base 
trace is, of course, determined by the values of L % and C 2 forming the 
tuned circuit of the oscillator. The time constant C 3 /? 2 should be 
large compared with the duration of one cycle of the oscillator frequency. 
The time constant C 5 J? 6 should be about one-twentieth of the duration 
of one cycle of the oscillator frequency, 



OH.T.t- 




Synch 



Fig. 114. 



■OOut 



o//.z- 



The circuit of a quenched calibrating oscillator which provides a 
pulse-form output 



It will be apparent that, when using a quenched oscillator of the 
type described above, no specific frequency relationship between the 
oscillator and the time base is necessary. In fact, the oscillator fre- 
quency may be altered, by adjustment of C 2 , in order to make two of 
the timing pulses coincide with centimetre marks on the measuring 
rule at any desired part of the trace. 

Such a method is, however, quite unsuitable for use with a time 
base in which the degree of linearity is really good since the error in 
such a trace would be too small to be detected in this way. Moreover, 
the accuracy of the deflection for a given potential applied to the 
deflector plates of a cathode ray tube will probably be little better than 
1 per cent and the method does not permit this error to be separated 
from the error in the potential wave-form. One might ask why it is 
necessary to make a time base with this degree of linearity if the error 
cannot readily be detected. The answer to this question is that for 
normal cathode ray tube applications it is unnecessary for the time 












LINEARIZATION OF THE TRACE 147 

base trace to be linear to within less than 2 or 3 per cent, since this 
will result in a displacement error which is much less than the error 
to be expected in the cathode ray tube itself. It is, however, sometimes 
desirable to apply a time base to a circuit without providing a visual 
representation and, in this case, greater precision of linearity may, with 
advantage, be employed. Many examples of circuits which require 
extreme accuracy are to be found in radar equipments, in particular, 
in those used for accurate short-range measurements for gunnery 



QHX+ 




1 Rz 

Control %„ 
Potential > K i 



Control 
Potential 

rARi- 1 - — o-5/os 



Fig. 115. Laws' pulsed calibrating oscillator 

purposes and for long-distance bombing and navigating equipments.* 
Circuits suitable for these purposes are described in Chapter IX. 

Laws' Pulsed, Calibrating Oscillator 

Lawsf has developed a pulsed oscillator whose frequency can be 
accurately adjusted to that of a standard crystal generator. This circuit, 
which appears in Fig. 115, has a negligible starting delay, reaches its 
full amplitude in the first half-cycle, and has the amplitude maintained 
at this constant level by means of a feed-back system, Changes in the 
potential of the H.T. rail of 1 per cent result in a frequency change of 
4 parts in 10 6 while the temperature coefficient has been made of such 
a value that a change of i°F. results in a frequency change of 3-3 parts 
in io 6 . 

The valve V 1 normally passes anode current, which is suddenly cut 

* Various authors. Journ. I.E.E., 1946, 93, Part IIIA; also Radiation Laboratory 
Series, McGraw Hill. _. 

+, C. A. Laws: "A Precision-ranging Equipment using a Crystal Oscillator as a 
Timing Standard," Journ. I.E.E., 1946, 93, Part IIIA, p. 423- 



148 



TIME BASES 



off by the application of a negative pulse to its grid. The resulting 
train of oscillations, which are damped by the resistances R 5 and i? 6 , 
are fed via C 4 to the grid of the cathode-follower buffer stage V 2 from 
whose cathode it is passed to the grid of V s . This valve applies positive 
feed-back to the oscillating tuned circuit L 2 C 2 via the coil L v The 
suppressor grid of V 3 is switched by means of a wave-form which is 
the reverse of that applied to the grid of V x . Thus, when V 1 is cut off 
to start the oscillations, V 3 is switched on to apply positive feed-back 



IfflffllMfflnpilJF 

A BRIGHT CRYSTAL- 

SPOTS CALIBRATOR 
PIPS 

iiiiiiiiiiiiiiiiiiiiiiiimr 

B 

TMMMM1MI1T 

C 

Fig. 116. Visual method of setting the frequency of Law's pulsed oscillator 

(A) Frequency is correct. 

(B) Frequency much too low or much too high. 

(C) Frequency slightly too low. 

and so to build up the oscillations quickly and to control their amplitude 
at a constant level. Control of the degree of feed-back is obtained by 
adjustment of the resistance i? n . When V t is switched on, the feed- 
back is removed because the suppressor grid of V 3 is again cut off. 
The oscillations are then rapidly reduced in amplitude by the resistances 
R s and R e . 

The frequency of the oscillator is adjusted by an approximate adjust- 
ment of Z, 2 , followed by adjustment of C 2 . A trimmer condenser (not 
shown) may be connected in parallel with C 2 for fine adjustment, and 
a temperature-compensating condenser may also be shunted across C 2 . 
The cathode-follower valve V 2 is used, on account of its high input 
impedance, to avoid too great loading of the oscillating circuit. 

The method adopted by Laws for ensuring that the oscillator fre- 
quency is correct is of considerable interest, since it provides alignment 
to within 1 part in 50,000. A crystal oscillator, whose output is 
squared and differentiated to provide calibration pips, has its output 
wave-form applied to the Y deflector plates of a cathode ray tube. The 
pips are suppressed at intervals to form trains of waves and are also 






LINEARIZATION OF THE TRACE 



149 









employed to trigger the time base and the pulsed oscillator. Each cycle 
of the output of the pulsed oscillator, after being squared and differen- 
tiated, is applied to the grid or cathode of the cathode ray tube to 
provide a series of brightening pulses. If the pulsed oscillator is 
operating at the same frequency as the crystal calibrator the image on 
the cathode ray tube will appear as shown in Fig. 116 (a), while if the 
frequency is much too low or much too high it will appear as in (&), 
and if slightly too low, as in (c). 

If the oscillator is operating at a frequency which is slightly too high 
it will appear as in Fig. 1 16 (c) but with the line of spots sloping upwards 
i nstead of downwards for the same direction of the time base traverse. 

Since 1 mm. measured along the leading edge of each crystal cali- 
brator pip in Laws' apparatus represents an error of 1 part in 50,000, 
and as it is possible to observe an error of less than 1 mm. in the parallel 
spacing of the trace and the bright spots produced by the quenched 
oscillator, the setting accuracy is better than 1 part in 50,000. 

The Trace Shift Method of Determining Linearity 

The trace shift method of determining linearity is carried into effect by 
shifting the trace so that the calibrator pulses are brought successively 
behind a line drawn across the face of the tube. The shift potential 
required to move the trace from one calibrator pulse to the next is 
measured in each case and the accuracy of the method depends on the 
accuracy with which the pulses can be set on the line and with which 
the various values of the steady shift potential can be determined. One 
factor which may be important if the time base output impedance is 
high, is the absence of deflector plate current; this can be avoided by 
ensuring that the mean potential of the deflector plates never becomes 
positive with respect to the final anode of the cathode ray tube. This 
condition may be ensured by employing the circuit of Fig. 117 provided 




b\ hH + 



Final Anode 



Fig. 1 17. The trace shift method of determining linearity 



IS© 



TIME BASES 



that the time base potential is arranged to drive the deflector plate 
increasingly negative as the trace proceeds. 

It is important to note that the commencement of a saw-tooth wave- 
form is always curved (see Appendix IV, page 343), and this portion 
should therefore be neglected in checking the linearity. 

Method of Determining the Linearity by Measuring the Slope of 
the Flux Variations 

Seeley and Kimball* have described a method of determining the 
linearity of an electro-magnetically produced time base trace by 
measurement of the rate of change of the flux variations. 

If a small pick-up coil is placed as near as possible to the centre of 

the coil system, a potential e=k — will appear across the unloaded 

at 

terminals of the coil. In this equation, ^ is the instantaneous flux and 

k is a constant which depends upon the number of turns in the pick-up 

coil and its position with respect to the field coils. 

The potential appearing at the pick-up coil terminals is thus propor- 
tional to the rate of change of flux and this is, of course, proportional 
to spot velocity when the pick-up coil is replaced by a cathode ray tube. 

The usefulness of the method lies in the fact that the differentiation 
of the flux variations provides changes which, when amplified and 
examined on a separate cathode ray tube," make the errors much more 
readily discernible than is the case when calibrating marks are used to 
check the linearity. 

Fig. 1 18 shows, at the top, a saw-tooth wave-form which is perfectly 
linear: the fly-back is also linear. Below this is shown the time derivative, 
the dotted line representing the average value which is also the line of 
zero spot velocity. 



Flux 
Wave- 
form 




.{_ . -3verage_J^aJue Line 

4_ . J^'mwg~Pp'/£ 



Pick-up j 
Potential 1 , 




Fig. 118. The flux and pick-up 

coil wave-forms for a linear trace 

having a linear fly-back 

* S. W. Seeley and C. N. Kimball: "A New Method for Determining Sweep 
Linearity," R.C.A. Review, 1940, 4, p. 338. 



Fio, tig. The flux and pick-up 

wave- forms for a non-linear trace 

having a linear fly-back 



LINEARIZATION OF THE TRACE 



Isl 



Fig. 119 shows the same two curves for a non-linear trace in which 
the fly-back is, for convenience, shown as linear. 

Examination of these two figures will show that the time differential 
curves (the lower curves in the figures) can more conveniently be 
employed if the large peaks, due to the fly-back, are clipped off so that 
the important part of the curve can be amplified until it fills the greater 
part of the screen of the cathode ray tube which is being used for the 
measurement. It is important, however, to know where the wave-form 



1 — o+h'.r. 




Output 



-HT. 



Fig. 120. Seeley and Kimball's circuit for determining the linearity of the trace by 
measuring the slope of the flux variations in an electromagnetic deflection system 

is being clipped because it is necessary to know where the "average 
value" line resides in the oscillogram. It is also important that this 
average value line shall remain in the same position on the cathode ray 
tube screen when the pick-up coil is moved or when adjustments are 
made to the time base deflection system which is under examination. 

A circuit for carrying out these measurements for a television raster 
appears in Fig. 120. When using the circuit for measuring the linearity 
of the horizontal trace, the vertical trace system should be rendered 
inoperative and vice versa. The pick-up coils which should be about 
\ inch or less in diameter have inductances of 10 microhenries for the 
horizontal and 2 millihenries for the vertical trace. These may con- 
veniently be fitted into paxolin formers which will just slide into the 
deflection coil system, so that the polarity and coupling may be altered 
by rotation of the paxolin former in the deflection system. The high- 
frequency coil (10 microhenries) should be of single-layer construction 
while the low-frequency coil may be a multi-layer coil. 

A peak-to-peak potential of approximately 0-5 volt is normally 
obtained and this should be applied to the input terminals in such a 
manner that the pulses due to the fly-back drive the grid of the valve 



152 



TIME BASES 



V x in a positive direction. In order to avoid wandering of the average 
value line rectification must not be allowed to occur in this valve, and 
the meter, in the circuit of Fig. 120, is employed to ensure that there is 
no change in anode current when the pick-up coil is placed in position 
in the deflection coil system which must be operating normally. 

The valve V x provides an output potential of some 6 to 8 volts 
peak-to-peak and this is applied to the grid of the second valve K 2 , via 
the condenser C 2 , in order to remove the steady component of potential. 
This valve F 2 is arranged so that the bias potential appearing at its 
grid is controllable by adjustment of the potentiometer R n for the 
purpose of setting the clipping-point. This clipping-point is also shown 
in Fig. 1 1 8. That part of the wave-form which appears in Figs. 1 18 and 
1 19, below the clipping-point, does not appear in the output of V 2 since 
it lies beyond the anode-current cut-off point. Since the steady current 
component of the wave-form appearing at the grid of F 2 is removed 
by C 2 , and the grid potential does not approach that at which grid 
current commences, the clipping-point is determined solely by the bias 
potential applied to the grid of V 2 . 

The value of the zero-signal anode current in V 2 , multiplied by the 
value of the anode load resistance R 1Z , is the potential difference 
between the clipping-point and the "average value" line on the cathode 
ray tube. Furthermore, if the output from the anode of V 2 is directly 
connected to a Y deflector plate of the cathode ray tube employed for 
measuring the rate of change of flux, the Y shift control may be 
adjusted to bring the average value line (zero signal line) into coinci- 
dence with a horizontal line drawn across the face of the tube. Then, 
provided neither the shift potential nor the bias potential at the grid 
of V % are allowed to alter, the average value line will remain fixed in 
position. Periodical checks may be made by removing or rotating the 
pick-up coil so that no input signal exists and by checking that the 
anode current of V % remains constant. A D.P.D.T. meter switch is 
provided in the circuit to enable the meter to be changed over to read 
the anode currents of V x and V 2 . When reading the anode current 
of V x , a meter shunt is automatically inserted to increase the full- 
scale current reading of the meter from i-o milliampere to 100 
milliamperes. 

Curvature of the characteristic of the valve V 2 tends to exaggerate 
any errors in linearity by making the time derivative of the flux wave- 
form appear to be more curved than it really is. This is, if anything, 
an advantage. 

Grid bias potential adjustment is provided for the valve V x in order 
to permit a certain amount of gain control, but care must be taken, as 
already stated, by checking the anode current with and without an 






LINEARIZATION OF THE TRACE 153 

input signal, to see that no change of anode current and, hence, no 
rectification occurs. 

The frequency- and phase- response of the amplifier must be suffi- 
ciently good to enable good square- wave transmission, at both the line 
and frame frequencies, to be achieved. This is a most important point 
since failure to achieve this object will introduce errors which may be 
mistaken for non-existent lack of linearity. Non-linearity of the trace, 
as seen in the television or other viewing tube, of course, appears as a 
lack of parallelism between the upper part of the time derivative curve 
and the average value line on the measuring tube. 

Any non-linearity of the time base trace in the cathode ray tube 
employed for measuring purposes is, in general, likely to increase the 
apparent error in the non-linearity of the increase of flux in the electro- 
magnetic deflection coil system since, normally, the rate of traverse of 
the spot on the measuring tube and the rate of change of flux in the 
deflection coil system will both be reduced towards the end of the trace. 
The time base trace on the measuring tube should occur once for each 
reversal of flux on the deflection coil system, and it is therefore advisable 
to synchronize the measuring tube trace from the valve system which 
drives the deflection coils. 

It may be necessary to add further amplification depending upon the 
sensitivity of the cathode ray tube employed for measurement. In this 
case, it is advisable to employ a steady-current amplifier and to ensure 
that no rectification occurs by checking that no change of anode 
current occurs upon removing the signal. Direct connection to the 
measuring oscillograph tube should also be employed. The require- 
ments regarding frequency- and phase-response also hold for any 
additional amplifying stages which may be added. 

The following component values may be employed in connection 
with the circuit of Fig. 120: 



*i 
R, 

R* 
^10 



2,500 ohms 

1 megohm 

200 ohms 

as required 

2,000 ohms 

4,000 ohms 

15,000 ohms 

25,000 ohms 

1 megohm 

200 ohms 



Kit 

R\t 

Rn 

c a 
c, 

c 4 
c 6 
v x 

V, 



25,000 ohms 
1,000 ohms 
250,000 ohms 
4,000 ohms 
025 u-V 
0-25 [aF 
0-25 t^F 
0-5 nF 
0-25 nF 
SP42 
6SJ7. 



Methods of Measuring the Linearity of a Raster 

When two linear time base potentials are applied at right-angles to 



154 



TIME BASES 



one another by connecting them to the two pairs of deflector plates 
of a cathode ray tube, the resultant image is known as a raster. For 
television reception, the two time bases must be synchronized and 
there must be a predetermined relationship between the two repetition 
frequencies. The form of the raster developed by the Marconi-E.M.I. 
Company and adopted in England by the B.B.C, provides 202^ hori- 
zontal lines for each vertical traverse. The complete raster pattern is 
traced out during each pair of frames* and it therefore contains 405 
lines. The lines provided during alternate frames are displaced verti- 
cally by half the distance between the lines of one frame so that the 
alternate 202^ line scans are interlaced. The method of building up 
the raster employed by the B.B.C. is shown in Fig. 2. The time base 
which produces the horizontal scanning lines is known as the line time 
base, while the other one is called the frame time base. 

The linearity of the time bases used in a television receiver may 
readily be verified by observation of a test pattern radiated for the 
purpose of tuning the receiver but, if no suitable signal is available, 
various devices may be employed to produce a pattern on the screen 
for the purpose of determining the linearity. A very rough idea of the 
linearity of the vertical traverse is afforded by observation of the 
illumination of the raster, since the screen will appear brighter where 
the lines are crowded. 

In order to measure the linearity of the vertical traverse more accur- 
ately, the line time base may be replaced by another operating at a lower 
repetition frequency, but with approximately the same amplitude, so 
that the individual lines are clearly distinguishable due to their greater 
separation. The horizontal traces then serve as calibrating marks and 
render the employment of a calibrating oscillator superfluous. The 
frequency of the horizontal sweep should be about thirty or forty times 
the frequency of the frame time base and the synchronizing terminal 
of the latter should be connected to an appropriate point in the circuit 
of the line time base. The linearity of the horizontal deflection is, in 
this case, of no importance since it is only required to obtain an 
indication of the variation in the vertical displacement between suc- 
cessive traces. Measurements may be made by means of a vertical 
scale which should be aligned with the centre of the screen in order 
to avoid the effects of barrel, pin-cushion, or trapezium distortion 
(see page 279). 

An alternative method, which does not involve alteration of the line 
time base, is to apply a modulating signal to the grid or cathode of the 
cathode ray tube, at least the frame time base and the modulating signal 
oscillator being synchronized with one another. The modulating signal, 

* In England, one frame is one vertical traverse. 



LINEARIZATION OF THE TRACE 



■>:> 



the frequency of which should again be about thirty or forty times the 
frame frequency, may conveniently be provided by a simple oscillator 
or time base. Horizontal black bars are produced on the raster and 
non-linearity of the vertical sweep is indicated by their uneven spacing. 
If the frequency of the modulating signal is raised to a multiple of 
the repetition frequency of the line time base, the observed effect is 
the appearance of alternate black and white vertical stripes. The 
uniformity of their spacing now indicates the linearity of the horizontal 
sweep which may be measured by means of a horizontal scale. In 
order to obtain a stationary pattern, it is necessary to synchronize the 
line time base from the modulating signal, or vice versa, although it is 
immaterial whether the vertical sweep is synchronized or free-running. 
According to one scheme, this difficulty is avoided with the aid of 
an oscillator which is quenched during each fly-back period so that 
the modulation of successive horizontal traces is aligned, although the 
frequency of the modulation may not be a whole multiple of the line 
frequency (see Fig. 114, page 146). 

In making measurements, the distance from the leading edge of one 
black bar to the leading edge of the next is taken; it is not advisable 
to measure the width of one bar because this depends upon the ampli- 
tude of the modulating signal unless the signal wave-form is perfectly 
square, which is unlikely to be the case. 

The modulating signal may be obtained from any type of oscillator 
which provides an output of 20 volts or more, at a frequency in the 
region of thirty to forty times that of the line frequency, although a 
square- wave oscillator is preferable in order to provide sharp edges on 
the test pattern. On the other hand, the possibility of frequency modula- 
tion of the oscillator at the line frequency is a serious danger and, for 
this reason, it is probably better to use a beat frequency oscillator and 
to synchronize the line time base from it. Frequency modulation of the 
modulating oscillator at the line frequency, or at any multiple thereof, 
is indistinguishable from velocity distortion of the horizontal deflection. 
In the Marconi-E.M.I. television transmissions emitted from the 
Alexandra Palace by the B.B.C, the maximum period of time allowed 
for the fly-back on the receiver is rinrWo of a second, which is approxi- 
mately equal to 10 microseconds, while the total trace time, including 
the fly-back period, is Tvkss of a second, or approximately 100 micro- 
seconds. Therefore, if 25 vertical black bars are required, the frequency 
of oscillation must be approximately 250 kilocycles per second. 

In practice, it will prove almost impossible to quench and restart 
an oscillator whose period is 4 microseconds (250 kilocycles per second) 
in a period of 10 microseconds without the existence of random phase 
changes. The required decay and build-up times may, however, easily 



156 



TIME BASES 






LINEARIZATION OF THE TRACE 



i57 



be achieved by employing two oscillators having frequencies of, say, 
30 and 30-25 megacycles per second. The resultant beat note is 
employed for brightness modulation and one or both of the oscillators 
may be quenched during the fly- back period. 

A further method of measuring the linearity of a raster or of a time 
base trace is given on page 279, and another method has been described 
by Duke.* 



oH.T.-h 




off.T- 



Fig. 121. Circuit for measuring the characteristics of a 
current-regulating circuit 

Method of Measuring the Characteristics of a Current Regulat- 
ing Circuit 

Since the rate of charge of a condenser is proportional to the current 
flowing into it, the linearity of a capacitive time base is readily deter- 
mined by means of steady-state measurements of the current-potential 
characteristics of the current-regulating circuit. This provides a useful 
verification of the most critical part of the time base, although it does 
not take account of other possible sources of non-linear distortion. 

The procedure to be adopted must be decided according to the 
circumstances of each case and several pitfalls are possible. The 
general idea may be seen from Fig. 121, which illustrates a case in which 
the required information cannot readily be abstracted from published 
valve characteristics. The time base shown here consists of a con- 
denser C x which is discharged by a trigger circuit, denoted by the key 
K t , and charged by the pentode, whose operation is made more linear 

* V. J. Duke: "A Method and Equipment for Checking Television Scanning 
Linearity," R.C.A. Review, 1941, 6, p. 190. 



by the introduction of negative feed-back. The potentials of the screen 
and grid are determined by the potential- dividing network R 7 R S R 9 , 
and the charging current is controlled by i? 4 (see page 39). The 
resistances R B and R G are stopper resistances of low value inserted to 
prevent parasitic oscillation. 

In order to derive the current-potential characteristics of a circuit 
of this type, it is necessary to commence by measuring, with the aid 
of the cathode ray tube used with the equipment, the maximum and 
minimum values of the anode potential excursion relative to the 
negative rail with the time base in normal operation. At the same time, 
the potentials at the points S and G should be measured with a high- 
resistance D.C. meter. 

After these preliminaries, the time base circuit is broken at the 
point X and the measuring circuit is applied via the dotted connections. 
The chain of resistances marked R x may well consist of ten 2,500 ohm 
1 watt resistances in series, the end units being variable to permit 
fine control of the applied potential, coarse control being obtained by 
selection of a suitable tapping point on the chain. In this way, the 
potential across the voltmeter V may be set to various values in order 
to plot a curve, or the currents at the two limiting values of anode 
potential may alone be measured. The anode current is measured by 
adjusting R% until no deflection is obtained on the centre-zero galvano- 
meter G, when the key K 2 is closed. R 2 preferably consists of a decade 
resistance box, although an ordinary wire-wound variable resistance in 
series with a much larger fixed resistance may be employed, since it is 
only necessary to vary R z over a small range. The resistance jR 3 serves 
to control the sensitivity of the galvanometer, while B Y is an accumu- 
lator or dry battery. The anode current is given by the potential of 
B x divided by the value of R % when the galvanometer is balanced. 
Since it is required to know only the ratio of maximum and minimum 
currents, it is not essential to measure the potential of B x accurately. 

During the normal operation of the time base, the grid and screen 
potentials are held constant by the condensers C 2 and C 3 but, during 
the steady-state measurements, the variation of screen current tends 
to alter these potentials and to produce a false change of anode current. 
This difficulty is avoided by connecting the batteries B z and B 3 as 
shown, in order to maintain the previously measured potentials at the 
points S and G. 

It is to be noted that the feed-back resistance i? 4 controls both the 
linearity and the mean value of the charging current. The method, 
therefore, permits one to measure the linearity in relation to the value 
of R i7 the value of the anode current being kept constant by adjustment 
of the potential of the battery B 3 . 



CHAPTER IX 
MILLER-CAPACITANCE TIME BASES 

The trigger circuits described in this chapter* differ from those dis- 
cussed in Chapter V in that they employ an effect known as the 
"Millerf Effect." This effect makes use of negative feed-back for 
increasing the linearity of the trace while positive feed-back is employed 
to operate the trigger portion of the circuit. 

General Theory 

Closely related to the circuits of Figs. 98 to 100 inclusive, there is a 
family of time bases which utilize the device known as a Miller- 
Capacitance Valve, i.e. a valve circuit with a condenser C connected 
directly between the grid and the anode. The insertion of this condenser 
results in feed-back from the anode of the valve to its grid. When such 
a valve circuit is provided with a resistive anode load the effect on the 
grid circuit is substantially the same as if a capacitance of C (A-\-i) in 
series with a resistance equal to \jg had been connected between the 
grid and the cathode, where A is the gain of the valve stage and g is the 
mutual conductance in amperes per volt. This is usually known as the 
"Miller Effect." 

The successful application of this principle to the production of time 
bases having a high degree of linearity is due to the following causes : 

(a) The use of a high degree of negative feed-back of the form 
described results in the grid potential being kept within the limits of 
the grid base of the valve and thus any change of grid potential is kept 
small with respect to the anode potential changes. The higher the gain 
of the circuit, the smaller is the resultant necessary change of grid 
potential during the linear part of the working stroke, i.e. the "run- 
down"! period. The change of grid potential is equal to the change in 
anode potential divided by the amplification factor for the valve in 

* The reader is advised to make himself familiar with the Miller-Capacitance Inte- 
grator, described on page 340, before proceeding to read this chapter. 

f J. M. Miller: "Dependence of the Input Impedance of a Three-electrode Vacuum 
Tube upon the Load in the Plate Circuit," Sci. Papers of the Bureau of Standards, 
1919, 15, p. 367; F. M. Colebrook: "A Generalized Analysis of the Triode Valve 
Equivalent Network," Journ. I.E.E., 1928, 66, p. 157; C. J. Mitchell: "Miller Effect 
Simplified," Electronic Engineering, June 1944, 17, p. 19; F. C. Williams: "Intro- 
duction to Circuit Techniques for Radiolocation," Journ. J.E.E., 1946, 93, Part IIIA, 
p. 289; F. C. Williams and N. F. Moody: "Ranging Circuits, Linear Time Bases and 
Associated Circuits," Journ. I.E.E., 1946, 93, Part IIIA, p. 1189. 

{The run-down period is defined as that part of the anode potential wave-form 
which occurs when the anode potential is decreasing in a linear manner. It is some- 
times preceded by a sudden fall in anode potential. 

158 



MILLER- CAPACITANCE TIME BASES 



i59 



the non-feed-back condition. For an EF50 valve, the amplification 
factor can easily reach a value of 1,000 or more. 

(b) The resulting high degree of constancy of the anode current 
results in extreme linearity. 

(c) A high degree of amplification with a very low output impedance 
is realizable in practice. 

It will be realized that (a) and (b) above mean that the valve functions 
as an extremely efficient current regulating device, while (c) means that 
it possesses the valuable advantage of a cathode follower, viz. low output 
impedance without the corresponding disadvantage that the output 
impedance can become high as is the case when the ordinary cathode 
follower has its anode current cut off. 

The principle of operation is basically the same as described earlier 
with reference to Figs. 98 to 101 inclusive, except that the cathode 



;*x 



bsm 



1 — ° £ ^ 

> Output 



&- 



V, 



+-o£ { - 




oEi+ 



Fig. 122. Basic circuit of a constant- 
current time base of the Miller- 
capacitance type 



°£j- 



Fig, 123. Fig. 122 redrawn to show 
the currents flowing in the circuit 



potential is held fixed instead of that of the anode. In the case of Fig. 105 
inspection of the diagram will show that the condenser C\ is connected 
between the anode and the grid of the valve. 

The basic circuit of a Miller-Capacitance Time Base is shown in 
Fig. 122 and this should be compared with Fig. 150, which is derived 
from Fig. 98. In this form, Fig. 122, the valve is provided with an infinite 
anode load except for the path through Cj and R v The switch shown 
in Fig. 122 is to be taken to represent any method of charging the 
condenser. It may be noted in passing that the condenser becomes 
charged when the switch is closed because the grid potential is 
limited by the flow of grid current and hence the lower plate of the 
condenser C x is held at the potential of the negative rail during the 
charging operation. While the switch is open, the anode draws a steady 
current through the condenser and the resistance, the latter having a 
value high enough to produce a negative bias at the grid. In order to 
make this point clear, the circuit of Fig. 122 is redrawn in Fig. 123 and 
the direction of the currents through the various components during 



i6o 



TIME BASES 



the run-down period is indicated by the arrows. If the current is to 
remain constant, the grid potential must change as the anode potential 
changes but, obviously, it cannot go to cut-off and must remain within 
the grid base. Likewise, it cannot go to such a value as will cause grid 
current to flow since this current would have to come from the con- 
denser and would result in a change in anode current. Hence the grid 
potential must be negative. The current through the resistance R t is 

E 
thus approximately * and this current flows via C x and the anode of 

the valve. The condenser is thus discharged at a rate given approxi- 
mately by — -j- volts per second, R x being in megohms and C x in 

microfarads. The negative bias at any instant during the discharge is 
equal to the difference between the supply potential E x and the product 
of the current and R v As the condenser discharges, the anode potential 
falls and the valve requires a reduced grid bias in order to maintain 
the same anode current. Because i? 1 is large, an appreciable decrease 
in grid bias can be obtained for a trifling reduction in the current 
through R x and the net result is that the current in R lt which is the 
same as the anode current, only falls very slightly as the condenser 
discharges. If the valve characteristics were perfectly linear over the 
operating range concerned, the potential rise at the grid, however 
slight, would be in constant proportion to the fall of the anode potential, 
and therefore the potential at the anode would be strictly proportional 
to that across the condenser. However, as the anode falls to the region 
of the knee of the pentode characteristic, the grid potential begins to 
rise more rapidly, the discharge rate of the condenser decreases and 
the anode potential becomes distorted in comparison with that existing 
across the condenser. Finally, the grid potential settles in the neigh- 
bourhood of zero and the remaining fall of the anode potential is very 
non-linear. Normally, one arranges matters so that the condenser will 
be recharged before the distortion becomes serious. A trigger device, 
for instance, can be set to produce the fly-back when the anode potential 
falls to, say, 50 volts. 

At very high trace velocities (order of 2,000 volts per microsecond) 
the time base is liable to be affected by the presence of stray capacitances, 
although it is possible to determine from purely steady-state measure- 
ments what the performance at low frequencies will be. By way of 
example, Table III (page 161) was obtained by connecting a battery 
across C t in Fig. 122 with the positive end of the battery joined to 
the anode end of the condenser, and measuring only the changes in 
the current through R x for various battery potentials. The supply 
potential E t was 300 volts, R x was approximately 100,000 ohms and 



MILLER-CAPACITANCE TIME BASES 



161 



3 





H 


— 71 
1 


T 


+ 


m 

ON 


I 


O 

n 

1 


1 


l 


O 

O 

m 




— Tl 
OO 

1 


U"4 

N 
+ 


6\ 


1 


1 


' 


I 


O 

CI 


M 

1 


\0 

1 


m 

+ 


m 

00 


1 


1 


1 


I 


O 

0~ 
•n 


H 

1 


— 71 


M 

4- 




ON 


1 


1 


1 


i 


O 

-if* 
~& 



T 


— 71 
l 


m 

+ 


to 

00 

ON 


in 
vO 

On 
On 


M 
i 


00 

O 

+ 


00 

6 






M 

~o 

»n 

M 


1 


1 


M 
+ 


0» 


OO 

On 

* 

hi 


H 

t 


O 

N 

O 

+ 


NO 

O 

6 
in 


O 

M 

O" 
O 

M 


1 


M 

1 


O 
+ 


m 
On 


On 
ON 
On 


Mf4l 

1 


in 

O 
+ 




NO 

H 
O 

6 


M 


O 
O 

N 

O 
in 


1 


— 71 

l 


in 



+ 


10 

ON 

On 


ON 

On 


1 


M 
O 

9 

+ 


M 

Th 
O 
O 

6 


Change of potential across 
C 2 in volts . 


< 

a 

.5 
3 

1 

5 

a> 
S 

H 

3 




a 

1 
w 

a 

D 

3 
a 

"w 
*-> 
O 

H 


■*-> 

c • 
S 

+-• 



a, 

r2 

"C 

w 



&8 
11 


c • 

"2 

a ■ 

*-> 

a, 
a> 

O 
§ 

«** 



bJQ +2 

§ g 
U 


■ 
§ 

■4-> 
O 

a. 
a 

*0 



a 

1*4 

Z 8 

a! ^ 

a -3 

8 -2 
«-• 
H 


.£ e 

S c 
c-S 1 

a! M fcj 

■-> ■-> — 


u S 
« 

B 

.tJ 

3 en 
D 

£ W 
s 

C u 

S * 


DO 

■M 

*3 

3 

a 
c 



c 

9 

c 

J 



13 






1 62 



TIME BASES 



the initial current for 250 volts applied across the condenser was 
approximately 3-0 milliamperes. The corresponding measured grid 
potential was —4-2 volts. The remaining data was calculated from the 
measured current changes. 

Normally, one takes the percentage change in current as being the 
linearity error of a time base (see page 113) but, in the present instance, 
this figure is not of much value because it is not the condenser potential 
which appears across the output terminals. Nevertheless, the figure is 
of interest. Since the current was 3-0 milliamperes at 250 volts and has 
fallen by a total of 3-5 microamperes at 50 volts, the linearity error of the 
condenser discharge for the range 250/50 volts is 0-117 P er cent. 

The change of grid potential from the initial —4-2 volts is derived 
from the change in current (in microamperes) by multiplying it by 
o-i (the resistance R t being 01 megohm). Note that, as the condenser 
discharges, the grid bias decreases and the grid potential moves in a 
positive direction. The change of anode potential is therefore equal to 
the change in the condenser potential minus the change in grid potential. 

The relative time taken for each 50 volts of condenser discharge is 
based on an arbitrary figure of 50 as the time that would be required 
if the current remained constant at the initial 3-0 milliamperes. The 
time is therefore given by 

(Initial currentX 
Mean current/ 

The time taken for the condenser to discharge from 250 volts to 150 
is thus 50-0041 +50-0125 = 100-0166, and, similarly, the time to reach 
50 volts is 200-0874. Considering the discharge from 250 volts to 50 
volts, the timing accuracy of the condenser potential at half-sweep is 
given by 

Time to 150 volts 100-0166 

Half of time to 50 volts 100-0437 ^"973* 

i.e. the timing error is 0-027 P er cent. 

Now consider the output potential at the anode; it will be seen that 
the anode potential falls by 99-9 volts (i.e. 49-95 +49-95) in the time 
100-0166 and 199-65 volts in the time 200-0874. At half the total time, 
i.e. 100-0437, tne correct output potential is half of 199-65, i.e. 99-825 
volts below the initial potential. From this, the timing error at the 
anode at half-sweep is 



99-9 - 99825 
99-825 



X 100= 0-076 per cent. 



The error at the output terminals is thus about three times as great 
as the error incurred in the process of discharging the condenser. It 



MILLER-CAPACITANCE TIME BASES 



163 



may be noted that the ratio of the grid potential change to the anode 
potential change gives an approximate measure of the amplification 
factor of the valve. In this case, the highest value indicated is about 
1,000, although pentodes are available with amplification factors in the 
region of 3,000 or more. The accuracy of the time base depends on 
both the amplification factor and the mutual conductance. With some 
high-gain H.F. pentodes the current is so nearly constant that it is not 
easy to measure the lack of linearity. 

The theoretical performance may be considered as follows : 
In the region above the "knee," the anode current characteristics 
may be approximately represented by an equation of the form 

where i a is the instantaneous anode current, 
e g is the instantaneous grid potential, 
e a is the instantaneous anode potential, 
g is the mutual conductance in amps, per volt (fi /#„), 
i? a is the internal anode resistance, and 
I Q is a constant. 

The value of 7 can be obtained by measuring the anode current of 
the valve at zero grid bias and with the highest anode potential it is 
proposed to use. The measured current will be greater than /„ by the 

amount — . In making calculations of the gain of an amplifier, the I 

term is usually ignored, because it does not contribute to the amplifi- 
cation process and is generally blocked off from the grid of the next 
stage by means of a transformer or a coupling condenser, but it is not 
negligible in the present case. 

In the absence of grid current and currents flowing into stray capaci- 
tances, the whole of the anode current passes through C l and R lf 
so that 

% = £1 - 4*i. -■- (2) 

If the potential across the condenser is denoted by e c , then, during the 
run-down, 

*« = e c + e r ... (3) 

Inserting (3) in (1) 

e c +e g 



4 = h +g e s + 



*« 



■4 + ( ! fK 



1 64 TIME BASES 

Eliminating e R , using (z) t we have 

Collecting terms in i a and denoting ' by £' 



R. 



This may be rearranged 



e 1 + y + - 



i« = 



£ I* H- i 



if 



(4) 



E 

It will be seen that the anode current is not widely different from — l , 

R i 
In order to minimize the effect of the e e term on the current, one requires 

only a valve having a sufficiently high amplification factor. The value 

of i?j has no direct effect on the relative influence of e c . The current 

is also liable to change as a result of variation in the value of g' which 

tends to fall when e c falls, and this effect may be the more serious of 

the two. 

By inserting (4) in (2) we get 



\ g V- +i / 



-< -I 








M£'*i + ^-g'R^E: 




n — 


■) 


*'*i H ' 








E -I R -Y*l* 








*'*i + 1 








_([x + 1) (*,-!«*,)- /- 


Ri«, 






fr+i)(l'*x + i) 






Substituting this in (3) gives 








f ^(^ + l)W+lh +(!* + 


0(^i 


-/o^i) 


- A''*l*c 



(1* + !)(*'*! + 1) 



MILLER-CAPACITANCE TIME BASES 



165 



(y. + i)(g / * 1 + i)-g , *i 



^1 






A\ - / n A\ 



I — 



\L + I 



tf„ -r 



g'Rx + I 



£. — 



£'* 1 + 1 

Mi - Si 



■(5) 



This equation (5) applies as soon as the switch is opened. 
If J R 1 > ljg' this reduces to 



3 V + 1/ £'*! 



which is an approximation which will, in most cases, give a sufficiently 
close answer. 

The first term is the major component of the anode potential and 
it will be distorted if the value of p changes, even if it changes linearly 
with e a , although the actual magnitude of the resultant "bowing" of 
the saw-tooth sweep may be quite small, especially if the lowest value 
to which \i falls is of the order of some hundreds. The second term is 
of the order of a few volts only. Since it varies inversely with g\ non- 
linear variation of ljg' will cause distortion although a change of ijg' 
which is linear with respect to e a would be harmless. 

During the operation of the circuit as a saw-tooth generator, equation 
(5) applies as soon as the switch is opened. At this instant the con- 
denser is fully charged and e t =E V The value of e a given by (5) for 
this condition is less than Ep mainly on account of the second term, 
and the anode potential falls until this equation is satisfied. This process 
occupies only a short time whose duration is determined by the stray 
capacitances and it is usually completed before C x has discharged 
appreciably. In effect, the anode and grid potentials move together in 
a negative direction for a few volts until a state of equilibrium is reached 
and then the linear sweep commences, with the anode potential falling 
and the grid potential rising slightly. 

The linearity of this time base appears to be somewhat inferior to 
that obtainable with the same components arranged as in Fig. 97 (on 
page 1 29), but it possesses the great practical advantage that, owing to 
the low output impedance, an appreciable load may be connected across 
the output without producing serious non-linearity. This load may 
consist of either a resistance or a capacitance or of both in parallel. In 
the case of Fig. 97, the output is taken directly across the condenser 



i66 



TIME BASES 



terminals so that any resistance load will, in effect, destroy the constant- 
current charging feature, since it will appear as a shunt to the effective 
charging resistance, which is of the order of several hundred megohms. 
If it were practicable to take the output between the anode and cathode 
in Fig. 97, without leakage to any part of the H.T. supply , there would 
be no difference in the performance of the two circuits. 

In the circuit of Fig. 122 the effect of a capacitance C 2 connected 
between the anode and cathode is that the anode continues to draw 
substantially the same current through C ± but, in addition, it has to 
accept the discharge current of C 2 . Both of these currents are nearly 
constant and the total anode current is similarly nearly constant. The 
linearity is only indirectly affected, the chief effect being that the grid 
potential during the sweep is less negative than it would otherwise be, 
and it may be necessary to employ a higher value for R l7 together with 
a reduced value for C l7 so as to keep the grid sufficiently negative to 
avoid grid current. Assuming that this point is attended to and pro- 
viding that C x is not made so small as to become comparable with the 
stray capacitance C g existing between the grid and cathode, it is not 
impracticable to load the anode with a capacitance C 2 many times the 
value of C v The fly-back remains good with the switching arrangement 
shown, but the circuit takes appreciable time to settle down afterwards 
so that the commencement of the sweep is non-linear. Normally, of 
course, one would desire to keep the value of C 2 as low as possible. 

It is of interest to consider the effect of stray grid-cathode capacitance. 
First, an increase of C g increases the amount by which the anode 
potential falls before the commencement of the linear run-down. Even 
if C g =C l7 however, the reduction in the potential range of the run- 
down is very small because the initial sudden fall of potential is only 
a very small part of the total linear excursion. Second, a part of the 
current through R x goes to charge C g during the linear potential fall 
because the grid potential is then rising. This effect is also very small. 

A Practical Circuit 

Several variations of this time base (see, for example, Fig. 124) involve 
the addition of a resistance R 2 between the anode and the positive H.T. 
rail with the result that the anode has to take a steadily increasing 
current as the anode potential falls. At the beginning of the sweep, the 

E 

anode current is approximately equal to - 1 and, if the sweep is termi- 
ni 

nated at an anode potential E 0t the current will then be approximately 

E t + E l -B 

/tj R 2 



MILLER-CAPACITANCE TIME BASES 



167 






If R» were connected between the anode and cathode, the current 



would rise from —^ — 



E t 



to — - approximately. 



Ri R% 
In each case, the change of current during the sweep is the same, i.e. 



Ei 



R t 



'°, and, in fact, there will be the same change no matter what 



the value of the steady potential applied in completing the circuit 
between R 2 and the anode and cathode. Thus, a resistance which is 
isolated from the time base by a blocking condenser will have the same 
ability to disturb the anode current, and 
it should be considered as being in 
parallel with any other anode load re- 
sistance when estimating the extent of 
the current change. When calculating 
the actual limits of the anode current, in 
order to be sure that it does not approach 
zero at the beginning of the sweep and 
does not become high enough at the 
other extreme to cause the grid to con- 
duct, the mean potential across the blocking condenser must be 
estimated. If this potential is E Mf the maximum current will be 

E i _j_ e m — E o 



o£ 2 + 



o Output 
-o Input 



«4- 



Fig. 124. A practical form of the 
Miller-capacitance time base 




*l 



lir 



The mutual conductance of a pentode, in general, increases with the 
anode current, more or less in direct proportion, but it is comparatively 
insensitive to changes of anode potential until the potential falls to the 
region of the "knee." The addition of the resistance R 2 complicates 
the situation by providing a large change of anode current. Since the 
anode potential is falling while the current is rising there is a tendency 
for the mutual conductance to decrease at both ends of the sweep, but 
the resulting distortion is usually insignificant if the value of R 2 is not 
made too low and R x is appropriate. 

The linearity of the run-down is proportional to the stage gain and 
to the potential applied to R x and inversely proportional to the potential 
change at the anode or grid during the run-down. Normally, only the 
potential applied to R x and the stage gain are controllable, since the 
trace potential is generally determined by the sensitivity of the cathode 
ray tube. The gain may be increased by choice of valve or by increasing 
the value of R 2 . If, however, R 2 is made too high in value, the valve 
will draw a reduced anode current which will reduce the value of g. 
Since the gain equals gR%, there is an optimum value for R z . 



1 68 



TIME BASES 



The choice of a suitable value for R ± need not be a matter of trial 

and error since it is not difficult to estimate the maximum current that 

the valve will take at a grid bias of, say, i to 1-5 volts and E anode 

E E 
volts. The current through R 2 , corresponding to E , i.e. — i- ° } 

should be deducted from this and the remainder taken as the current 
to be obtained through R x at 2?! +1-5 volts. When the circuit contains 
a capacitance C 2 connected between the anode and cathode, the current 

C 

remaining after allowing for R 2 should be multiplied by 1 to 

obtain the fraction which will be drawn through R v The value of R x 
calculated in this way will be the minimum value which can safely be 
used without risk of grid current. If the calculated minimum is already 
higher than the value of R z it should be taken as an indication that R 2 
is excessively low and a valve capable of passing more current should 
be employed. 

When this circuit is incorporated in a time base intended to cover 
a wide range of repetition frequency, the coarse adjustment of sweep 
velocity is generally obtained by switching in different condensers while 
a fine adjustment is provided by varying the value of R l or the potential 
applied thereto. 

If the condensers are in steps of 10 : 1, the current will have to be 
made variable over a range of about 11 : i, and it is preferable to 
select capacitances in steps of 1, 2, 5, 10, etc., so that a 3 : 1 variation 
of R t will suffice. 

Since the trace velocity is almost directly proportional to E lt this 
being the potential applied to R lt the repetition frequency can be made 
insensitive to ordinary changes in an unstabilized H.T. power supply 
if the trigger fly-back device is designed to operate when the anode 
potential falls to a definite fraction of E x rather than to a fixed potential. 
Thus, if E x rises by 10 per cent, the rate of discharge will be increased 
by 10 per cent and the duration of the discharge down to any given 
fraction of the starting potential will be unchanged. 

For some purposes it is necessary to connect the resistance R x to a 
potential supply which is independent of the initial potential at the 
anode at the beginning of the sweep. The duration of the sweep is 
then substantially proportional to this initial potential of the anode 
and inversely proportional to the potential applied to R x , 

With a trigger device which is suitable for single-stroke operation, 
the fly-back at the end of the sweep can be made to produce a pulse 
which is delayed by a controllable time interval after the initiating 
signal, regardless of whether the signal recurs at regular or random 
intervals. 



MILLER-CAPACITANCE TIME BASES 



169 



Values of C x as small as 5 micromicrofarads can, with care, be used 
and the resulting time base trace duration is excessively short. A grid 
stopper resistance of approximately 500 ohms should be used with all 
Miller-capacitance time base circuits. 

The following values are suggested for the circuit components: 

R x 100,000 ohms 
R 2 300,000 ohms 
R 3 50,000 ohms 
R 4 50,000 ohms 
R b 500 ohms 
H.T. supply 300 volts 
Valve EF50. 

Reversed Operation 

If the resistance R x is connected between the grid and a negative 
supply potential, the anode potential will rise during the sweep while 
the grid becomes more negative. The switch is then arranged to bring 
the anode potential back to a starting-point in the region of 50 volts, 
or less, and a diode is connected to the grid to limit the negative 
movement of the grid potential so that the condenser C 1 can be dis- 
charged. For the normal operation of the circuit, the lower end of the 
condenser is held substantially at the potential of the negative supply 
rail by grid current. In the reversed method of operation, this does 
not occur and, hence, a diode is necessary. The cathode of the diode 
is connected to the grid of V\ and its anode is connected to a point 
which is preferably a few volts negative with respect to the cut-off 
potential of V v Alternatively, the condenser may be discharged by 
driving the grid positive until the anode potential falls to the desired 
level. 

With this mode of operation, the presence of R 2 is essential and it 
must have a value at least several times smaller than R x since, otherwise, 
the anode current becomes cut off before the anode potential has risen 
to a large fraction of the available positive H.T. potential. A large 
inductance may be connected in series with the anode load resistance 
to reduce the change of anode current and allow a larger sweep amplitude 
to be obtained. 



External Operation 

A time base of the type shown in Fig. 125 is set into action by the 
removal of the cut-off bias on the suppressor grid as shown in Fig. 126. 
The anode potential rapidly falls a few volts until the grid has reached 
an equilibrium condition and then falls steadily to the "knee," after 



iyo 



TIME BASES 




Fig. i2S- 



which it becomes non-linear. The restoration of cut-off bias allows 
the anode potential to rise exponentially and, during the intervals 
between sweeps, the grid current is limited by R x while the screen 
current is limited by R z . This time base is chiefly of use in cases where 
there is a relatively long blacked-out waiting period between sweeps in 
which case the slow fly-back is of no consequence. 

It is advisable to feed the triggering 
square wave to the suppressor grid via a 
condenser with a leak resistance to earth 

f H \ o Output and a diode V z across the resistance so 

\Ci - - ; o Input that the suppressor cannot be driven 

positive. This arrangement is shown in 

Fig. 127. If desired, the resistance i? 4 

may be omitted and a square- wave pulse 

can then be obtained from the screen 

grid. The rate and, hence, the duration 

of the linear fall of potential at the 

anode can be controlled by varying the potential to which the upper 

end of the resistance R x is connected and this results in a variation in 

width (duration) of the pulse obtained from the screen grid. 

Occasionally, a diode V z is connected between the anode of the valve 
and a potentiometer connected across the supply rail with a suitably 

1 — oH.T-f 

Output 
Saw-tooth 

> Trace 
"j ' ~*% Amplitude 
s Control 

o Output 
Square-wave 

o Input 



0£ 2 + 



**r 



The Miller-capacitance 
time base 




Suppressor Volts 



Fig, 126. 
Wave-forms pertaining to the 
circuit of Fig. 125 



* — oH.T- 



FlG. 127, The circuit of Fig. 125 modified to provide 
controlling circuits and both saw-tooth and square- 
wave output potentials 



large capacitance connected between the potentiometer slider and the 
negative rail. The anode of the diode is connected to the anode of the 
pentode. Variation of the potentiometer setting adjusts the level from 
which the fall of anode potential commences and, since the rate of fall 
is unaltered, the duration of the run-down and, hence, of the square 
wave are also varied. A number of circuits based on these principles 



MILLER-CAPACITANCE TIME BASES 



171 



have been developed by Newsam, Blumlein, Whiteley, Williams and 
others.* 

Attree's Time Base 

Attreef has described a method of reducing the time occupied by the 
fly-back. In his circuit, the condenser C x is not directly connected to 
the anode of V x but is joined to the cathode of an additional triode 
whose grid is connected to the anode of V v The triode anode is con- 
nected to the positive supply rail and a resistance of high value is 
connected from the cathode of the triode to a source of negative 
potential. During the trace, or run-down period, the triode valve behaves 
as a cathode follower so that the condenser C x is effectively connected 
to the anode of V v During the fly-back period, however, the increase 
of potential at the anode of V x drives the grid of the triode into the 
grid current condition and the condenser is then rapidly charged by the 
anode current of the triode. 

An increase of velocity of the fly-back stroke, up to about 100 times, 
can be achieved, depending on the value of the ratio of the anode load 
resistance R 2 to the internal impedance R„ of the triode when grid 
current is flowing. Thus, a triode which is capable of passing a high 
value of anode current should be used if the maximum reduction of 
fly-back time is required. 

It will be appreciated that the circuit will be most effective at the 
lower frequencies, where the ratio CJC S is high, C s representing the 
stray capacitance between the anode of V x and the negative supply rail. 

MacKay's Time Base 

It is sometimes required to delay the commencement of the run- 
down and an artifice may be employed to provide a result of this nature. 
MacKayJ has developed a circuit to do this. 

A resistance and a capacitance, in series, are connected in parallel 
with the condenser C x in Fig. 127, the condenser being connected to 
the junction of C x and R v A diode has its cathode connected to the 
junction of the additional resistance and capacitance and its anode is 
connected to a potential which is, say, 50 volts less than that of the 
positive supply rail. 

The additional resistance is not of sufficient value to reduce appreci- 
ably the current in the added capacitance and, therefore, the Miller 

*B. Newsam: British Patent 493843; J. W. Whiteley: British Patent 57525°; 
A. D. Blumlein: British Patent 580527; J. M. Debski: British Patent 580891 ; F. C. 
Williams: British Patents 582758 and 584329; F. C. Williams and N. F. Moody: 
British Patent 587364. , , 

t V. Attree: "Improving Fly-back time on a Miller Time Base," Electronic Engineer- 
ing, 1948, 20, p. 97. 

{ D. M, MacKay: "A Multiple-Pulse Generator for Synchronized Transmitter 
Systems," Jottrn. I.E.E., 1946, 93, Part IIIA, p. 1199. 



172 



TIME BASES 



time constant is increased and the run-down takes place more slowly, 
i.e. its slope is reduced. 

When the anode potential has fallen by 50 volts, the diode passes 
anode current and fixes the potential of the junction point of the added 
components. The Miller time constant is then reduced to its original 
value and the remainder of the run-down takes place at the normal rate. 

The circuit does not, in fact, delay the commencement of the run- 
down but it has a similar effect which is produced at the expense of a 
reduction in the maximum normal- velocity run-down potential. 

THE TRANSITRON MILLER TIME BASE (PHANTASTRON). 

There are a large number of Miller-capacitance time bases which 
include a transitron circuit arrangement. This group of time bases is 
generally known by the name Phantastron,* which is sometimes 
abbreviated to "Phant" or "Fant." In all these circuits, the time 
constant is increased and the linearity improved by the employment of 
the Miller effect, while the transitron portion of the circuit is used to 
provide the necessary feed-back which is required to cause the capaci- 
tance to discharge after the linear part of the run-down. Of course, 
when the charging or, as it is sometimes called, the integrating resis- 
tance is connected to a source of potential which is negative with respect 
to the ordinary negative supply rail, the run-down becomes a run-up 
and the condenser is then afterwards charged by the transitron action 
of the circuit. 

As already described in Chapter V, the transitron may be screen- 
coupled, in which case the suppressor-grid potential varies with that 
of the screen because they are coupled together via a condenser, or it 
may be cathode-coupled, in which case the suppressor grid is held at 
a fixed potential relative to the negative supply rail and, owing to the 
presence of a cathode load resistance which is not by-passed, the 
suppressor grid is cut off when the cathode potential rises by a pre- 
determined amount positive with respect to the negative rail due to 
the passage of screen current through the cathode load. These two 
types of transitron circuit arrangement lead to a terminological separa- 
tion of the phantastron into two groups as follows: 

(a) Cathode-coupled Phantastron (C-Phant). 

(b) Screen-coupled Phantastron (S-Phant). 

Furthermore, the phantastron can be triggered by a positive or a 
negative initiating pulse by making a small modification to the input 
circuit. The initiation of the action in a phant is brought about by 
causing the anode to pass current and this may be done by applying 

• The name phantastron alludes to the fact that the circuit is a transitron whose 
performance was, at the time of its invention, considered (rightly so) as fantastic. 



MILLER-CAPACITANCE TIME BASES 173 

a positive potential to the suppressor grid in a screen-coupled phant or 
by lowering the potential of the cathode and, hence, effectively raising 
that of the suppressor grid by making the grid potential more negative 
in a cathode-coupled phant. 

The Cathode-coupled Phantastron 

The cathode- coupled phantastron has one stable and one unstable 
limiting condition and may be used in two ways : 

(a) as an accurate delay circuit, 

(b) as a divider. 

The phantastron is not very useful as a time base since the anode 
potential falls by about 30 volts from its stable value before it com- 
mences the linear run-down portion. Other circuits have been devised 
in which this potential fall is only about 3 volts, e.g. the screen-coupled 
phantastron (see page 178). Moreover, the fly-back is very slow 
and, therefore, it is only useful for examining the wave-form of signals 
which occur at intervals which are long compared with their duration. 
This phantastron is, however, very useful for producing a signal 
having a specified, and sometimes variable, delay with respect to the 
initiating pulse and for frequency or time division. These functions are 
intimately related to those performed by a time base. 

One of the most notable facts about these Miller circuits, and 
especially about those circuits which have been derived from them, is 
their apparent complexity — in some cases as many as five or six valves 
are used — but most of these valves are diodes which are cheap and 
occupy little space, while the advantages which accrue from their use 
are considerable. 

The circuit of a cathode-coupled phant* is shown in Fig. 128 and 
the appropriate wave-forms appear in Fig. 129. It will be seen that 
there are several potentials existing in the quiescent condition of the 
circuit and these should be fairly closely adhered to for efficient operation 
of the circuit. These potentials are as follows: 

(a) Cathode of F 3 -}- 41 volts. 

(b) Suppressor grid of V z + 20 volts. 

(c) Cathode of V 2 + 37 volts. 

(d) H.T. supply rail + 300 volts. 

The control potential which is required for operating this circuit is 
a negative pulse of about 25 volts amplitude having a sharp leading 
edge and this is applied to the cathode of the diode V % via the con- 
densers C x and C 2 and the resistance R v 

* F. C. Williams: British Patent 582758. 



i 7 4 



TIME BASES 



Since the anode of the diode V 2 is connected via the resistance R 7 
to the positive H.T. supply rail, the diode normally conducts anode 
current and the resistances R ly R &i i? e and R 7 are adjusted to make the 
cathode of V 2 and, hence, the grid of V 3 reside at a potential of 37 volts 
positive with respect to the negative supply rail. The potential of the 
cathode of V 3r which has a short suppressor-grid base (VR116 or 
Mazda V872), is adjusted to reside at 41 volts positive by adjustment 
of R t . Likewise, by adjustment of R iy the potential of the suppressor 
grid is made to be 20 volts positive. The values given in the table of 



OH.T.+ 




Control 
Potential 



Fig. 128. A cathode-coupled phantastron 

suggested values for the components on page 178 should be suitable 
for providing these required potentials if the H.T. supply is maintained 
at 300 volts and adjustment of their values should not, in practice, be 
required. 

Since the potential of the cathode of V s is 41 volts and that of the 
suppressor grid is only 20 volts, the suppressor grid, whose potential 
is 21 volts negative with respect to the cathode, cuts off the anode 
current and the whole of the current flows to the screen grid in the 
static condition. Upon the application of the 25 -volt negative control 
pulse, the cathode of V 2 is driven negative, its anode and the grid of 
V 3 follow it, and so does the cathode of V 3 by virtue of the presence 
of the unshunted cathode load resistance R l Q . The cathode of V 3 thus 
has its potential changed to, say, 41 —25 = 16 volts so that the suppressor 
grid changes in potential from 21 volts negative with respect to the 
cathode to 4 volts positive and this immediately causes a flow of anode 
current and, at the same time, drives the screen potential positive owing 



MILLER-CAPACITANCE TIME BASES 



i7S 



to the resultant fall in screen current which has been flowing through 
the resistance R 12 . 

When anode current commences to flow in V 3) the anode potential 
necessarily falls and this fall is applied via the condenser C 4 to the 
grid of V 3 which acquires an equilibrium potential from which point 
it slowly becomes slightly more positive as the run-down continues, as 
already described on page 160 et seq. The function of the diode V x is 
as follows: the potential of its cathode is determined by the setting of 
the slider of the potentiometer R 3 and, since its anode is connected to 
the anode of the pentode F a , the adjustment of R 3 also determines the 
potential of this anode. The potential from which the anode of V 3 
commences its run-down, and hence the duration of the run-down, are 
therefore controlled by adjustment of R 3 . The duration of the approxi- 
mately square wave-forms provided at the screen grid and cathode of 
V 3 are likewise adjusted by the same operation. 

Examination of the wave-forms of Fig. 129 will show that the potential 
of the grid of V 3 reaches a value which is in the neighbourhood of zero 
volts before the linear part of the anode run-down commences so that 
the anode current in the diode V % is cut off even though the control 
pulse may not be completed. Since the diode V 2 now has no effect 
upon the operation of the circuit, the valve V 3 continues to function 
as a Miller time base until the end of the run-down is reached. The 
rate of fall of anode potential is very nearly 



dt 



E 

CJi 7 



where E is the potential of the H.T. supply rail. Note that the adjust- 
ment of the potentiometer R 3 alters the starting potential for the 
run-down, but has no effect upon the rate of run-down. Thus, suppose 
that C 4 has a capacitance of 30 micromicrofarads and that R 7 is 
2 megohms, with E equal to 300 volts, then the rate of run-down is 

300 x 10 _ vQ j ts p ef m i crosecon d p 

30 x io~ 12 x 2 x io 6 

When the anode potential approaches the completion of the run- 
down, i.e. when the anode potential approaches a potential near the 
knee of the anode volts-anode current characteristic, it remains fairly 
constant while the total space current increases since the grid becomes 
more positive. Although the space current increases, the anode current 
falls and that of the screen grid increases. Since the anode potential is 
relatively constant, the feed-back ceases, thus permitting the increase 
of grid potential mentioned above, together with an increase in the 



176 



TIME BASES 



cathode potential, by cathode-follower action until, finally, the cathode 
potential exceeds that of the suppressor grid whereupon the anode 
current is once more cut off. The anode and grid volts rise rapidly 
towards the potential of the H.T. supply rail until that of the grid is 
caught at 37 volts by the diode V % . Until the diode V z commences to 
conduct when the grid reaches a potential of 37 volts, both ends of the 
condenser C 4 move positive at the same rate. Thus, until the potential 

reaches 37 volts, the time constant is ( 7 9 J C n where C a repre- 

sents the sum of the anode to earth and grid to earth stray capacitances 
together with that of C 4 and the wiring to earth. After the grid is 
caught by V 2t the lower end of C 4 is held fixed and the anode potential 
rises more slowly because C 4 has to be charged via the anode load 
resistance R 9 . During this period, the time constant is i? 9 (C 4 -J- C as ), 
where C as represents the stray capacitance from anode to earth. There- 
fore, the rate of rise of the anode potential alone is considerably less 
than that of the anode and grid together because the resistance is 
slightly increased and the capacitance is greatly increased — about five 
times for C 4 equal to 30 micromicrofarads. When the anode of V s 
reaches the potential to which the slider of the potentiometer R 3 has 
been set, it is caught by the diode V ly whose impedance is low compared 
with that of the resistance R 9 , and held until the arrival of the next 
negative control pulse. 

An examination of Fig. 129 will show the various wave-forms in 
different parts of the circuit in response to the initiating control pulse 
shown in Fig. 129 (a). In particular, it should be noted that approxi- 
mately square wave-forms are obtainable from the screen grid, which 
provides a positive wave, and the cathode, which provides a negative 
one. If a differentiating circuit is connected to the screen grid or 
cathode output terminals, a pulse, which simulates that of the control 
pulse and has the same or the opposite initial phase respectively, can 
be obtained and this pulse will be delayed with respect to the control 
pulse by a time which is variable by adjustment of #3. The circuit can 
thus be used to provide a variable delay. 

Furthermore, examination of Fig. 129 (b) will show that the applica- 
tion of further initiating pulses after the grid potential has reached its 
most negative potential can have no effect until the completion of the 
run-down. If, therefore, control pulses continue to arrive at a rate which 
does not allow a second pulse to appear before the grid is fully negative 
or during the return stroke, the cathode-coupled phantastron can be 
used as a dividing stage. 

One of the disadvantages of the cathode-coupled phantastron is the 
large anode potential fall of about 30 volts which occurs before the 



T 



MILLER-CAPACITANCE TIME BASES 



177 



linear part of the run-down commences. This causes a time delay in 
the start of the effectively linear part of the time base trace, as seen on 
the cathode ray tube screen and, therefore, unless the control pulse 
occurs before the signal to be examined, thus bringing this signal on 
to the linear run-down, the latter will appear distorted. Another dis- 
advantage is that the cathode potential must change from 41 volts to 
approximately 2 volts, a total change of 39 volts. Since the cathode 

Delay T- 

\ s Input 

V 




Trigger 

Pulse 



Potential of 
Cathode of fy 

Potential of 
Grid of % 



Potential of 
Anode ofV 2 






Potential of 
Screen ofi§ 



Tic. 129. Wave-forms pertaining to the circuit of Fig. 128 

load resistance is 3,900 ohms, this involves a change of cathode current 
of 10 milliamperes. Therefore, the screen current during the stable 
regime must be 10 milliamperes greater than the sum of the anode and 
screen currents during the run-down period. The maximum anode 
current is, therefore, severely limited and, since this current passes 
through the charging or integrator resistance i? 7 , the rate of run-down 
for a given value of capacitance C 4 is also limited. This means that the 
cathode-coupled phantastron cannot be used for extremely fast rates of 
run-down. 

The following values for the components will be found to be 
suitable : 
M 



r 7 8 







TIME BASES 


*1 


22,000 ohms 




c, 


50-0 y-y-F. 


R 2 


15,000 ohms 




C 2 


o-i nF. 


* 3 


50,000 ohms 




***ft 


0-1 fiF. 


R* 


10,000 ohms 




C 4 say 30-0 iiy.F. 


*6 


2,700 ohms 




e* 


0-25 nF. 


*6 


5,600 ohms 




Pi 


VR92 


*7 


2-2 megohms 


v* 


VR92 


*8 


560 ohms 




v. 


VR116 or Mazda V872 


R* 


330,000 ohms 




H.T. 


300 volts (stabilized). 


^10 


3,900 ohms 








Ru 


5,600 ohms 








#12 


5,600 ohms 









It is generally advisable to stabilize the H.T. supply sources for 
Miller-capacitance time bases and 500-ohm grid stoppers are normally 
required. 

The cathode-coupled phantastron exists also in other forms, e.g. the 
initiating pulse may be applied to the suppressor grid of F 3 . 

The Screen-coupled Phantastron 

The screen-coupled phantastron has been developed, from the 
cathode-coupled version of the circuit, by Fleming- Williams.* The 
circuit is shown in Fig. 130, from which it will be noted that the control 
pulse is now applied to the suppressor grid directly instead of via the 
grid and cathode. Another notable difference which is, however, not 
visible by inspection of the figure, is that the cathode load resistance 
is reduced to 200 ohms but the circuit will, in fact, continue to operate 
if this resistance is reduced to zero. 

The potentials and currents which must exist during the stable 
regime are as follows: 



(a) Cathode potential 

(b) Screen potential 

(c) V 3 anode potential 



1-8 volts. 

100 volts. 

80 volts. 



Since the suppressor grid is initially biased beyond cut-off, no anode 
current flows during the stable regime and the anode potential is deter- 
mined by the setting of the slider of the potentiometer R lt because the 
diode V x conducts. The whole of the cathode current therefore flows 
to the screen grid and this current may be altered by changing the 
value of i? 5 , since the grid is held at earth potential by the diode F 2 . 

The control pulse, which must, of course, be positive, drives the 
suppressor and screen grids positive and initiates a flow of anode 
* B. C. Fleming-Williams : A. C. Cossor, Ltd., London, 



MILLER-CAPACITANCE TIME BASES 



179 



current. As a result, the anode potential falls, carrying the grid potential 
with it until an equilibrium potential is reached, whereupon the anode 
potential commences to fall in a linear manner during the run-down 
period. This equilibrium potential is that grid potential at which the 
valve passes sufficient current to ensure that a current equal to EjR 2 
flows through C 2 . The actual anode current during the run-down is 
E/Rz plus that which flows through R A , the latter, of course, increasing 
during the run-down. 

When the anode potential of V s commences to fall, the diodes V x 
and V % cease to conduct. The grid of V % is, at the same time, driven 



o#7> 




oOut 



Control 
Potential 

o 



Bias 
-t5V 



QH.T,- 



Fig. 130. The screen-coupled phantastron 



negative by approximately 3 volts and this is the amount by which 
the anode potential falls before the run-down commences. If the 
resistance R 5 has its value reduced to zero, this non-linear anode 
potential fall can be slightly reduced but, in this case, no output can 
be taken from the cathode — normally, however, the cathode output 
potential available with R B equal to 200 ohms is only about o-8 volt, due 
to a change of total cathode current from about 9 milliamperes in the 
stable, to about 5 milliamperes in the unstable condition. Obviously, 
if the cathode load resistance R 5 is increased in value, the output from 
the cathode can be increased, but this can only be done at the expense 
of an increased non-linear potential fall at the anode before the com- 
mencement of the linear run-down. 

When the unstable regime commences, the screen current falls to about 
3 milliamperes and, hence, the potential of the screen grid rises to about 
160 volts, which is maintained until the anode potential reaches its 



i8o 



TIME BASES 



fi 1 



lowest value at the end of the run-down period. At this time, the 
electric field between the anode and the suppressor grid has become 
too weak for the anode to draw through the suppressor grid all those 
electrons which have passed the screen grid. These remaining electrons 
then return to the screen grid, the potential of the latter therefore falls 
and carries that of the suppressor grid with it, so that the anode current 
is rapidly cut off. The potential of the anode therefore rises and carries 
that of the grid with it, both ends of the condenser C 2 rising in potential 
together until the grid potential becomes caught by the diode V 2 at the 
potential of the negative rail. Until this occurs, the time constant which 

controls the rate of rise of anode and of grid potential is ( - *^J- j 

where C s is the total stray capacitance of the anode, grid and wiring 
of the valve F 3 to the negative supply rail. After the grid potential is 
held by V 2 at zero potential, the condenser C 2 has to be charged and the 
time constant changes to R A (C 2 +C os ), where C a3 represents the total 
stray capacitance of the anode and wiring to the negative rail. The 
anode potential rises until it is caught by the diode V lt after which the 
circuit remains stable until it is once more triggered by the control 
pulse. 

Since the suppressor-grid potential is biased to —15 volts during 
the stable regime, a potential of at least 20 volts is required for the 
control triggering pulse, but this will only provide satisfactory results 
if the leading edge rises extremely rapidly. Fleming- Williams has found 
that a potential of about 80 volts provides the most satisfactory operation 
and results in output wave-forms which are most nearly square. The 
circuit is fairly flexible, however, and any potential above about 60 volts 
will produce good results. Moreover, fairly wide variations in the 
suppressor grid bias potential are permissible but 15 volts is a satis- 
factory value. 

When the control pulse triggers the circuit, the condenser C 4 becomes 
charged via the diode V A . Upon the completion of the pulse, C 4 dis- 
charges through i? 10 . The diode is thus left at a potential which is 
lower than 80 volts and until this potential has, at least partially, 
discharged through R 16i the circuit cannot be operated by another 
pulse unless it be of greater amplitude than the first. 

The linearity achieved with this circuit is likely to be about five 
times better than that obtainable with the cathode-coupled phant of 
Fig. 128 because, since the cathode load is reduced in value, a smaller 
change of grid potential is required to produce a given change of anode 
potential — the feed-back control of anode-current variation during the 
run-down period is therefore increased. 

The following components may be employed: 






MILLER-CAPACITANCE TIME BASES 



181 



*1 


25,000 ohms 


c t 


o-i t*F. 


*2 


500,000 ohms 


C 2 say 


ioo-o [j.{i.F. 


*3 


560 ohms 


C 3 say 


o-ooi ti.F. 


*4 


300,000 ohms 


c 4 


o-oi jxF. 


*5 


200 ohms 


c b 


0-25 {iF. 


R 6 


1 megohm 


v, 


VR92 


R 7 


10,000 ohms 


v 2 


VR92 


Rs 


10,000 ohms 


v 3 


VRn6or Mazda V872 


K 


150,000 ohms 


v, 


VR92 


R io 


100,000 ohms 


H.T. 


300 volts (stabilized). 



The only components which are likely to require alteration are R & 
and R x , which may, with advantage, be altered if the amplitude of the 
control pulse differs much from the values given in the text. 

TWO-VALVE MILLER TIME BASES 

The Sanatron 

The sanatron* is a development, by Williams and Moody, j of the 
phantastron. The sanatron circuit is one in which one valve performs 
the linear run-down and the other introduces the necessary change of 
feed-back conditions which are required to provide cyclic operation. 
This replaces the transitron effect in the phantastron. 

The Suppressor grid-controlled Sanatron 

The circuit of the suppressor grid-controlled sanatron is shown in 
Fig. 131, from which it will be seen that V 2 is the Miller-capacitance 
valve, the Miller condenser being C 3 while the integrator resistance is 
R 8 . The valve V n receives an input signal from the anode of V 2 via 
C 4 and R Q and feeds back a phase-reversed and amplified signal to the 
suppressor grid of V 2 via V 12 and R lQ . This valve, V 6 , is normally 
triggered via R tl and the diode F 4 , the triggering control potential 
appearing at the grid. The circuit may, however, be triggered by the 
following control potentials which must, in each case, be negative 
pulses : 

(a) 10 volts applied to the grid of F 6 via F 3 , 

(b) 60 volts applied to the suppressor grid of V G via a load resistance 
(not shown), 

(c) 20 volts applied to the cathode of V x via a cathode resistance 
(not shown). 

* The Royal Air Force used many slang terms. The name "Sanatron" has been 
derived from the term "sanitary," meaning satisfactory (see also page 250 et seq.). 
t F. C. Williams and N. F. Moody: British Patent 587364- 



l82 



TIME BASES 



When the circuit is in the stable condition, the pentode valves V 2 
and K 6 take grid current since both grids are connected to a 200-volt 
positive supply rail via the resistances R 8 and R 9 . The valve V 6 is in 
the condition known as " bottomed, "* and therefore the suppressor grid 
of V 2 is driven negative via R l2 and i? lp . When F 6 is bottomed, its 
anode potential is approximately 30 volts and the total potential across 
R 12 and R t is, therefore, 350 volts. The potential of the suppressor 
grid of F 2 , which takes no current, is approximately —70 volts for the 



o*300 V 



< I4 S"15 

-X-.-oOut (c) 
oOuttf) 

^~~>—oOut(b) 




o-4V 
O-300 V 



Fig. 131. The suppressor grid-controlled sanatron 



following reason. The sum of the resistances R l2 and 7^ 10 is 350,000 
ohms and the rate of fall of potential across them is, therefore, 1 volt 
per 1,000 ohms. Since R 12 is 100,000 ohms and the higher potential 
end thereof is at +30 volts, the lower end of R 12 and the suppressor 
grid which is connected thereto must reside at a potential of —70 volts 
if the suppressor grid takes no current, and this condition obtains at 
the potential referred to. The potential of the suppressor grid is pre- 
vented from rising above that of the cathode of the same valve, K 2 , by 
the diode V % . The condenser C 5 is placed in parallel with R 12 to 
provide a low-impedance path for the higher frequencies, and thus to 
avoid delaying the rapid rise of the amplified triggering control pulse 
supplied to the suppressor grid of V 2 . The resistance R 14 in the anode 
of V e is kept low in value for the same reason. 

Since the suppressor grid of V 2 is at a potential of 70 volts negative 

# The expression "bottomed" implies that the anode potential has reached a point 
on the anode volts-anode current characteristic curve below the knee, i.e. the anode 
potential is at the lowest value which it can assume with the existing circuit conditions. 



ft 



MILLER-CAPACITANCE TIME BASES 



183 



in the stable condition, the anode current in V 2 is zero and the potential 
at the lower end of i? 4 is determined by the setting of the slider of the 
potentiometer R 2 . Since no anode current flows in V 2 , the potential 
of the anode of this valve is also at the same potential as the anode of 
the diode V 1 and the screen grid takes the whole of the space current. 

Upon the application of the negative control pulse to the cathode of 
the diode F 4 , the grid of V Q is driven negative until, by the cumulative 
feed-back action of V 2 and V 6 , it is caught by the diode V 5 whose 
anode resides at a potential of —4 volts: V 9 thereupon has its anode 
current cut off and, as its anode load resistance is only 15,000 ohms, 
the anode potential rises rapidly and carries with it the suppressor grid 
of V 2 . Anode current thereupon commences to flow in V 2 and the 
anode potential of V 1 makes the normal non-linear fall, in this case 
about 6 volts, as explained later, and then proceeds to perform the 
linear part of the run-down, as explained on page 160, et seq. 

Owing to the presence of the resistance R 5 , the anode of the valve 
V 2 falls to a lower potential than that of the anode of the diode V-. so 
that the grid of F 6 is driven more negative than the grid of V 2 . The 
grid of V 2 is, of course, driven nearly to the cut-off potential by the 
feed-back action via the condenser C 3 . Therefore, the potential fall at 
the anode of V \ is given approximately by 

Cs j~ C * X (grid base of F 2 ), 

where C s is the stray capacitance between the grid of V 2 together with 
the wiring and the negative supply rail. This potential is of the order 
of 6 volts and, of course, this value applies only to the original non- 
linear excursion before the commencement of the run-down. The 
potential generated at the anode of V 2 to drive the grid of F s in a 
negative direction is now calculable. This will be 6 volts plus the 
potential generated in the resistance R 5 , i.e. 

6 -J- iR 5 volts. 



The current i in R^ is 



300 - E CV 1 200 



I: 



neglecting the effects of stray capacitances, where E CVi is the cathode 
potential of the diode V v The potential calculated from the equation is 

and is dependent upon the setting of the potentiometer R 2 . The 
potential drives the grid of F 6 negative until it is caught by the diode 
F 5 at —4 volts. F 6 therefore has its anode current cut off extremely 



-1F2- 



1 84 



TIME BASES 



rapidly owing to the large gain round the loop from the grid of F 6 to 
the suppressor grid of F 2 and back to the grid of F 6 , a cumulative 
action being introduced by the feed-back. 

During the run-down, a current -- . 



R, 



flows in Rf, and the rate of 



200 



run-down is ~~ w while the run-down commences at the potential to 

C'iRs 
which R 2 is set {E R2 ) and continues until the valve V 2 bottoms at, say, 

+ 3 volts. The duration of the run-down is, therefore, R2 3 8 

200 

seconds, where C 3 is in. microfarads and i? 8 in megohms. Since the 

rate of run-down is dependent upon the potential at the +2oo-volt 

supply rail, any variation in this potential during the run-down will 

cause non-linearity. This effect can be employed to alter the law of 

the run-down by cyclic variation of its value during the run-down 

period. As the anode of ^.bottoms, the feed-back via C 3 ceases and the 

screen-grid potential falls since, as the anode can take no more current, 

the extra current must flow to the screen and, therefore, the anode 

potential commences to rise. When this occurs, the grid of F G is driven 

positive and F 6 then drives the potential of the suppressor grid of V 2 

beyond cut-off, thus providing a cumulative increase in the positive 

pulse applied to the grid of F 6 . The circuit is then, once more, in the 

stable condition where It awaits another triggering control pulse. 

The wave-forms appearing at different parts of the circuit are shown 
in Fig. 132 and the letters at the side of the wave-forms also appear in 
the circuit of Fig. 1 3 1 so that they may easily be identified with the 
various output terminals. 

Recommended values for the components are given below: 



a, 


10,000 ohms 


&u 


15,000 ohms 


R* 


50,000 ohms 


*!• 


15,000 ohms 


R3 


12,000 ohms 


Rib 


2,000 ohms 


*4 


500,000 ohms 


c\ 


0-5 [lF. 


*5 


3,000 ohms 


c z 


0*5 nF. 


*. 


470 ohms 


c 3 


say 100 n^F. 


Rj 


470 ohms 


C4 


say 100 pf*F. 


R 8 


66,000 ohms 


c 5 


say 50 (itiF. 


R* 


150,000 ohms 


Vi 


VR92 


Ri* 


250,000 ohms 


7% 


VR91 (EF50) 


*n 


5,000 ohms 


v 3 


VR92 


^12 


100,000 ohms 


v. 


VR92 


R12 


470 ohms 


r* 


VR92 






v. 


VR91 (EF50) 



MILLER-CAPACITANCE TIME BASES 



185 



The suppressor grid-controlled sanatron can employ practically any 
pentode valve provided that the suppressor grid can be biased beyond 
cut-off with a potential of —70 volts. Moreover, much faster rates and 
greater linearity of run-down can be obtained than with the cathode- 
coupled phantastron, so that it is extremely valuable for use as a high- 
speed time base and it is remarkably satisfactory and flexible in 



Trigger \J\fov 



Anode V 2 




E 



(b) 



Time Constant 

R 2 (C 3 +C4 +Strays) 



._--+3f 



0-5 V (O 



GridV 2 ~— £—~- p r Cut-off 



Screen V e 



*-4V 

+JSOV 



(€) 



■+120V 



; Delay < O-ZS/* sec. 



Anode V 6 



+270 V 



r/; 



+30'/ 



Fig. 132. Wave-forms pertaining to Fig. 131 



operation. Faster rates of run-down are obtainable because the current 
through R 8 can be greater for the same applied potential than with the 
cathode-coupled phantastron and the anode current can also be greater. 
The practical limitation to the anode current for a given valve is that 
of the permissible screen wattage rating. During the stable period, the 
whole of the current passes to the screen grid, as has already been 
stated and, therefore, particularly when there is a long period between 
each operation of the circuit, the dissipation of power in the screen-grid 
circuit must be considered when choosing a valve to perform the 
function of V 2 - Moreover, it is important to ensure that the space 



1 86 



TIME BASES 



current during the stable period shall be several times as great as that 
during the unstable period since, otherwise, the run-down will be 
limited in amplitude because the anode potential will not approach zero 
so closely as is possible, at the end of the run-down period. This 
limitation may be avoided by controlling the flow of current in V 2 by 
alteration of the screen potential instead of by changing that of the 
suppressor grid, as explained later. 

The use of another diode and potentiometer circuit, similar to the 
arrangement of R lt R z and V 1 in Fig. 131, but with the additional diode 
having its anode connected to the additional potentiometer and its 
cathode to the anode of V v enables the potential of the upper end of 
the condenser C 3 to be predetermined at each end of the run-down. 
The duration of the run-down is then limited to the potential excursion 
E v\— E F2 > where E V1 and E V2 are the potentials applied bv the two 
potentiometers to the two diodes. In these conditions, a square wave- 
form, whose duration is very nearly 

T — ^V*8(jri~^Vj) 
200 
appears at the anode of V 6 . 

The Screen grid -controlled Sanatron 

In the screen grid-controlled sanatron there is no screen current 
during the stable period since the screen potential is arranged to 
alter from -36 volts during the stable, to +200 volts during the 
unstable period. In these conditions, since current flows in F 2 only 
during the unstable period, the occupancy ratio* is very low and the 
instantaneous power handled by the valve may be greatly increased. 
This ratio is of the greatest importance since it determines the value 
of the instantaneous current or power which the valve can handle. For 
example, suppose that a given valve can pass a maximum steady anode 
current of 10 milliamperes, if the occupancy ratio is 1/1,000, the current 
can be increased to 10 amperes provided that the cathode emission is 
sufficient. The maximum cathode emission of a 4-watt heater-cathode 
system of a normal receiving type of valve is of the order of 12 amperes 
When considering the question of power, the maximum steady power 
divided by the occupancy ratio is equal to the maximum available 
instantaneous power output. For this reason the values of the resistances 
R A and R 8 may be greatly reduced, the anode and integrating currents 
will increase and the rate of run-down will likewise increase. Rates 
as high as 100 volts per microsecond can be obtained with good linearity 
In trus form of the circuit, the function of V 2 can be performed by a 

* In a valve in which the space current is switched on and off, the time ratio on /off 
is known as the occupancy ratio. on/oft 






MILLER-CAPACITANCE TIME BASES 



187 



tetrode and, since these valves often have a higher power-handling 
capacity, it is sometimes advantageous to employ one. 

The Cathode-controlled Sanatron 

It is also possible to control the valve V 2 by raising the potential of 
the cathode above that of the grid during the stable period and a circuit 
providing this form of operation is shown in Fig. 133. In this circuit, in 
which the delay in the commencement of the run-down is reduced, the 



o+3QQ>/ 



O+200 V 




o+lO V 

Control 



Potential 



oOV 

o-4V 



Fig. 133. The cathode-controlled sanatron 






potential of the grid is fixed at about +10 volts by the diode F 5 and 
the valve V % conducts. V 9 must draw sufficient current during the 
stable regime to raise the potential of the cathode of V 2 to about 
+ 20 volts so as to cut off anode current in that valve. 

When a negative control pulse arrives, V & is cut off and the cathode 
potential of V 2 falls, so that anode current flows in that valve and the 
run-down commences. When the run-down is completed, V & is once 
more caused to conduct, owing to the trigger action via R 5 and C 4 , and 
these same components hold V a anode current cut off during the run- 
down period. The anode current of V z flowing through i? 6 also 
contributes to this effect. 

The Grid- controlled Sanatron 

Yet another method of controlling the valve V 2 is by switching the 
grid potential from a cut-off condition to one in which anode current 



z88 



TIME BASES 



flows. A circuit of this type, which is also due to Williams and Moody, 
is shown in Fig. 134. 

During the stable regime F 6 conducts, and its anode potential is 
low so that the potential of the cathode of F 3 , in the absence of the 
diode F 4 , is also low and V 2 anode current is, therefore, cut off. 

Upon the inception of a triggering control pulse at the cathode of 
the diode V lt the potential at the grid of V B falls, via the circuit R^C^, 
and anode current in that valve ceases, The cathode potential of V 3 



O+300V 




1 — 0+200 V 



MILLER-CAPACITANCE TIME BASES 



189 



oov 

0-3V 
O-6V 
-O -30OV 



Fig. 134. The grid-controlled sanatron 



rises and, when this potential and, hence, that of the grid of V 2 , passes 
the value for cut-off, anode current commences to flow in V 2 and the 
run-down begins. The diode F 4 is inserted in order to avoid the delay 
incurred while the condenser C 3 charges via the integrator resistance 
i? 6 to the point at which anode current commences. By connecting the 
anode of F 4 to a fixed potential of —3 volts, the grid potential of V 2 
is not allowed completely to cut off the flow of anode current in V 2 . 
Since the anode current in V 2 is not cut off, the value of the resistance 
i? 4 must be greatly reduced if the initial potential at the commencement 
of the run-down is not to be reduced, but, when this is done, it is 
likely that the final potential at the end of the run-down will be unneces- 
sarily high and the potential range of the run-down will be reduced. 
Moreover, i? 4 must be sufficiently low in value to permit the diode V x 
to conduct, even when R 2 is at its most positive setting since, otherwise, 






the control pulses cannot initiate a change of regime. The inductance 
Lj is inserted in the anode circuit of V 2 since, owing to its high impe- 
dance during the run-down, it permits the anode potential to fall to 
a very low potential without a large amount of anode current. The 
value of Lj may be made sufficiently large that i? 4 may be neglected 
during the run-down period and the additional current required to flow 
through i? 4 and L x during this period is inversely proportional to the 
value of Lj. 






PUSH-PULL DEFLECTION 



191 



CHAPTER X 
PUSH-PULL DEFLECTION 

The advantages to be gained by the use of push-pull for at least one 
pair of deflector plates are so great that unbalanced time bases are rarely 
employed. If it is essential to employ unbalanced deflection, it is 
advisable to use an anti-trapezium-distortion cathode ray tube (sec 
page 276). The time base in many commercial oscillographs is directly 
connected to one X plate and a single-valve phase-reversing stage, 
e.g. an amplifier having an overall unity gain with the output taken 
from the anode, is used to supply the other X plate. In certain cases, 
where the repetition frequency is low and a rapid fly-back is not 
essential, it is possible to employ a push-pull transformer, but this 
solution of the problem is not applicable when it is desired to vary the 
frequency over a wide range. The insertion of a valve amplifier 
between the time base and the cathode ray tube often involves some 
sacrifice in linearity and in fly-back velocity, although it is possible to 
arrange the circuit so that the distortion introduced by the valve 
approximately cancels out the effect of an exponential curvature in the 
wave-form of the time base. Unfortunately, variations in the gain of 
the phase-reversing stage are liable to occur and the deflector plates 
only receive symmetrical potentials if the circuit is kept in good adjust- 
ment. A small amount of unbalance of, say, 5 per cent to 10 per cent 
is not serious unless both the horizontal and vertical deflections are 
fairly large, but this error is not difficult to avoid if the amplifier is 
provided with negative feed-back to the extent of, say, 50 per cent to 
100 per cent. In these conditions it is well known that, although the 
gain is greatly reduced, it is influenced only to a small extent by wide 
variations in the characteristics of the valve while distortion due to 
curvature of the valve characteristics is likewise reduced. 

The amplifier may be used, as indicated above, to reverse the phase of 
the time base potential applied to one X plate of the cathode ray tube 
before application to the other. This method requires that the time 
base wave-form be generated at the potential level required for appli- 
cation to an X plate. 

An alternative method is to generate the time base potential at a 
low level and to amplify it by means of a circuit which provides a 
push-pull output of the required potential, the input being an un- 
balanced one. 

Five important methods of obtaining push-pull time base deflecting 
potentials are available. These are: 

190 






(a) A single-valve phase-reversing stage which may, or may not, 
provide more than unity gain. This may be subdivided into 
two types, viz. : 

(1) Those in which the input saw-tooth wave-form is sensibly 
linear and valve distortion is arranged to be a minimum. 

(2) Those in which the input wave-form is exponential and 
the amplifier characteristic is employed also for compensa- 
tion of the input curvature (see Bedford and Stevens' 
method of linearization on page 1 14 and Jenkins' method 
on page 1 16). 

(b) Single-valve amplifiers employing negative feed-back. These 
may be subdivided into two types, viz.: 

(1) Those employing current feed-back. 

(2) Those employing voltage feed-back. 

(c) Two-valve Push-pull amplifiers. These may be subdivided into 
four types, viz.: 

(1) Standard forms of push-pull amplifier. 

(2) Cathode-coupled push-pull amplifiers. 

(3) Anode-coupled push-pull amplifiers. 

(4) Amplifiers combining the principles of cathode and anode 
coupling. 

(d) Split time bases. 

(e) The use of two equivalent time bases operating in phase opposi- 
tion. 

It is debatable which of the above methods is the most satisfactory. 
Some are undoubtedly better than others as regards the reduction of 
distortion while others produce a greater output especially at the higher 
frequencies. Circumstances, however, alter cases and sometimes one 
method of obtaining push-pull potentials will be the most suitable and 
sometimes another. For instance, the Bedford and Stevens' method 
is recommended only when accuracy takes second place after expense. 
It is, however, possible to state without much fear of contradiction 
that (bi) the single- valve push-pull circuit employing negative current 
feed-back, where unity or less gain is required, (bz) those employing 
negative voltage feed-back, and (cz), (rj) and (^4) push-pull amplifiers, 
where amplification is required in addition to the provision of a push- 
pull output will, in general, provide the most satisfactory results. The 
other methods are described for the sake of completeness, but, obviously, 
the least distortion will be obtained if amplification can be entirely 
dispensed with since, in this case, the maximum degree of feed-back 
may be employed. Obviously, distortion may be still further reduced 
by employing (d) a split time base or (e) two equivalent time bases 



192 



TIME BASES 



operating in push-pull, thus avoiding the employment of any output 
valve stages. 

A real need for an amplifier arises in some cases, however, because 
there are many time base applications in which a small amount of 
distortion is acceptable and the inclusion of an amplifier permits special 
features to be obtained. As an example of this, the trace or a portion of 
it may be expanded by alteration of the gain of the amplifier without 
interfering with the operation of the time base. In this way a small 
portion of the trace may be examined in greater detail by operating the 
gain control. The cathode-coupled push-pull amplifier is of particular 
value in this respect, 

THE USE OF AMPLIFIERS FOR PROVIDING PUSH-PULL 

DEFLECTION 

(a) A Single- valve Phase-reversing Amplifier 
According to one point of view, the design of a time base amplifier 
should be based on a mathematical analysis of the wave-form in order 
to find the range of frequencies which must be transmitted, after which 
the ordinary principles of amplifier design may be applied to secure 
uniform amplification within this band. The wave-form is thus 
regarded as the resultant of a large, but not unlimited, number of 
harmonic components. If both the trace and the fly- back are linear, 
the highest order harmonic which need be taken into account is directly 
proportional to the period of the wave divided by the duration of the 
fly-back.* One may, for example, neglect all harmonics above the 
twentieth if the fly-back occupies at least one-twentieth of the complete 
cycle. Since a single valve is normally sufficient to provide any degree 
of amplification likely to be required for a time base, it is a simple 
matter to obtain uniform gain over a 20 : 1 frequency range, but it is 
much more important, and more difficult, to keep the phase shift 
sufficiently small. For instance, a phase shift of only i'5° at the funda- 
mental frequency will produce as much distortion as a 10 per cent 
variation in the charging current. Generally, phase shift at the lower 
frequencies is not a serious difficulty and is easily avoided by employing 
a large time constant for each resistance-capacitance coupling or de- 
coupling network. At the highest frequencies in the range, some phase 
shift is inevitably introduced by the stray capacitances, although various 
means of compensation may be adopted. 

In order fully to appreciate the real nature of the problem, another 
aspect should be taken into consideration. This is illustrated in Fig. 135 
with reference to a specific example. It is assumed that it is required 

•Manfred von Ardennc: "Distortion of Saw-Tooth Wave-Forms," Electronics, 
November 1937, 10, p. 36. 






PUSH-PULL DEFLECTION 



i93 



to produce a saw-tooth wave-form (a) having an amplitude of 200 volts 
and a linear fly-back the duration of which is 1 microsecond. If the 
period of the wave is twenty times that of the fly- back, as shown in the 
figure, the repetition frequency will be 50,000 times per second pro- 
vided that there is no waiting period between successive traces. 





Fig. 135. Compensating wave-forms relating to the design of 
amplifiers 

The wave-forms (b) and (c) are calculated for the case where the 
anode load resistance is 20,000 ohms (curve b) and the stray capacitance 
associated with the anode is 50 micromicrofarads (curve c). These are 
perhaps higher than necessary, but are convenient to illustrate the 
principle involved. A steady current of 10 milliamps is required to 
charge or discharge a capacitance of 50 micromicrofarads at the rate of 
200 volts per microsecond and 0-526 milliampere is required to charge 
it during the trace (19 microseconds).* The total anode current which 
has to be obtained is the sum of the currents (b) and (c) as shown 
at (d). 

•The appropriate equation will be found on page 305. 
15 



i 9 4 



TIME BASES 



PUSH-PULL DEFLECTION 



r 95 



If the valve is operated within the linear portion of its characteristic, 
the required grid potential will be a scaled-down copy of the wave- 
form (d). From this it becomes clear that the input wave-form to 
the amplifier valve should differ radically from the desired output 
wave-form and should consist of a brief pulse superimposed on the 
desired saw-tooth wave-form. During the intervals between pulses the 
input wave-form should be linear because the stray anode capacitance 
draws only a small constant current from the circuit during this period, 
assuming that the pulse discharges the capacitance to the correct 



OKT.+ 




o/f.r- 



Fig. 136. Time base employing a frequency-independent phase-reversing 
valve to provide a push-pull output potential 

potential at the beginning of each trace. If the amplitude or duration 
of the pulse is excessive, the observed effect will be that the fly-back 
overshoots the starting-point of the trace, as shown dotted on the 
wave-form (a). If, on the other hand, the pulse is of insufficient 
amplitude, the resulting curve of the output potential will be under- 
corrected. This condition is also shown dotted in the wave-form (a) 
in the figure. The exact shape of the pulse is unimportant since it 
is not usually worth while to attempt to linearize the fly-back. 

An amplifier circuit which makes use of this principle is shown in 
Fig, 136. The amplifier is intended primarily as a unity-gain phase- 
reversing stage for feeding the second X plate of a cathode ray tube, 
the other X plate being connected directly to the time base which is 
applied to the amplifier input. In order to obtain an overall gain of 
unity, the input potential is first scaled down by a network C^C^tJt^ 
which in the absence of R 3 is frequency-independent, the attenuation 
being adjusted to counterbalance the gain of the valve. The resistance 



R 3 , suggested by L. H. Bedford, is inserted to provide a brief positive 
pulse at each fly-back. When the input potential rises rapidly, a large 
potential is developed across R 3t driving the valve into grid current 
and discharging the stray capacitance C 6 , which includes that associated 
with the cathode ray tube. 

During the linear trace, the condensers C 1 and C 2 , in series, discharge 
through jR 3 as the input potential falls. The arrangement behaves as 
a differentiating circuit and a nearly constant potential is developed 
across R 3 during the trace. The value of jR 3 is best determined by 
trial but, as a rough guide, the time constant C. Z R 3 may be made equal 
to # 4 C 6 . 

In deciding upon the values required for the input network, the 
gain of the amplifier should first be determined (or calculated) without 
the network. The required attenuation of the network is then given by 

R x +R t A' 

where A is the gain as measured. Thus: 

R^R^A-i). 

Note that since the input terminals will normally be connected directly 
across a current regulating valve circuit D, the value of R 1 -\-R 2 should 
be high to avoid the introduction of serious non-linearity. 

If the constant current is to be, say, 2 milliamperes and the non- 
linearity introduced by the resistances is not to exceed, say, 1 per cent, 
the peak to peak value of the current which can be allowed to pass 
through the resistances is 0-02 milliamperes. If the maximum sweep 
potential is, say, 200 volts, the resistance R^R^ must not be less than 

200 , 

■ = 10 megohms. 

0-02 X 10" 3 

The required values for R t and R 2 can thus be evaluated. 

In order to determine the values of the condensers, 'C l should be 
made small, preferably not more than 20 micromicrofarads in order to 
avoid too great a reduction in the maximum possible repetition 
frequency of the time base. The value of the condenser C 2 is then found 
from the equation 

It is advisable to make C x adjustable since the value originally decided 
upon must, in practice, be increased by a small amount to compensate 
for the presence of stray capacitance between the grid and the cathode 
of the valve and because the capacitance of C 2 is unlikely to be of the 
precise value assigned to it. 



196 



TIME BASES 



It is also necessary to consider the effect of the condenser C 4 and 
resistance R 7 upon the wave-form. The time constant C 4J R, should 
be made at least 100 times the duration of the longest trace which 
the circuit is required to handle. In this case the non-linearity of the 
trace will be increased by a further 1 per cent. 

It is important to note that this scheme may fail to function properly 
if the fly-back is made to drive the grid negative, since the valve may 
then become cut off during the fly-back with the result that the stray 
capacitance C e cannot rapidly acquire the new potential. 

When the saw-tooth waves are separated by a waiting period, the 
problem of amplification is less difficult because the fly-back may be 
blacked out and the correction to be applied, to compensate for the 
presence of stray capacitance across the anode load, consists only of a 
constant potential added to the saw-tooth. This is demonstrated on 
page 193. Since the rate of change of potential during the trace is 
much less than during the fly-back, the distortion produced by the 
stray capacitance is comparatively small but the same method of 
compensation is applicable. If compensation is omitted, the effect 
of the stray capacitance is to round off the corners of the wave-form 
and to introduce a delay approximately equal to the time constant of 
the anode circuit. 

If it is desired to feed the amplifier directly from a condenser which 
is being supplied with a constant current, the compensation may be 
obtained by connecting a small resistance in series with the condenser C 
which is being charged, again making the time constant CR equal to 
the anode circuit time constant. 

The function of the compensating resistance in these circuits is to 
supply a potential which is proportional to the time derivative of the 
potential across the condenser. The method is thus identical in 
principle with the well-known method of compensating a resistance- 
coupled amplifier by inserting a small inductance in series with the 
anode load resistance, so that the inductance supplies a potential which 
is proportional to the time derivative of the potential across the anode 
load resistance. 

(b) Amplifiers Employing Negative Feed-back 

A Single-valve Push-pull Amplifier employing Negative Gurrent 
Feed-back 

Fig. 137 shows a simple push-pull stage based on the feed-back 

principle. The input potential is obtained from a time base which 

is represented diagrammatically by the condenser C x in series with a 

constant current device D. The key is understood to represent the 

fly-back device, which may be a gas discharge tube, blocking oscillator 

or any suitable hard valve trigger circuit. The anode and cathode load 






PUSH-PULL DEFLECTION 



197 



resistances R x and R 2 are made equal and, for optimum results, the stray 
capacitances of the anode and cathode circuits should be balanced by 
the addition of a small balancing condenser connected across R z or R i 
as found necessary. 

In the absence of grid current, the anode current is exactly equal 
to the cathode current and therefore the symmetry of the output is 
independent of any ordinary variations in the valve. Moreover, 
50 per cent of the output potential is fed back to the input so that 
distortion due to the valve is considerably reduced. In order to avoid 
grid current it is necessary to make the H.T, supply E x at least 100 volts 

-o+£ t 

o+E? 




vJf.T- 



F(G. 137. A single-valve push-pull amplifier employing negative current feed- 
back for use with a time base 

greater than twice the value of the potential E z . The potential across 
R 2 when the key is closed is obviously equal to E 2 if the grid-cathode 
potential is zero. If one allows, say, 100 volts for the anode-cathode 
potential drop and a further potential equal to E 2 for the drop across 
Ri it will be clear how the value of E l is derived. The resistances Rj 
and R 2 should be of the order of 50,000 to 100,000 ohms each, while 
R 3 and i? 4 may be 2 megohms each. C 2 and C 3 must be made large 
enough so that the time constants C 2 R Z and C Z R A are very large (say 
100 times) compared with the duration of one trace. If this is not so 
the trace will become non-linear, due to the reduction of potential across 
the condensers during the period of the trace. 

One great advantage of this type of circuit is that it can provide 
approximately double the output at high frequencies when compared 
with a valve whose output is taken only from the anode since the output 
is limited only by the necessity of avoiding the effects of stray capaci- 
tance across the load resistance. This applies particularly in those cases 
where the grid leak resistance and the output are both taken from the 
junction of a cathode bias resistor and the cathode load. In the case of 
a valve where the output is taken only from the anode however, there 



198 



TIME BASES 



PUSH-PULL DEFLECTION 



199 



is an additional output limitation due to the potential drop in the anode 
load, which results in non-linearity. In the push-pull circuit, on the 
other hand, the valve has a high value of internal impedance as explained 
on page 130, so that it behaves rather like a pentode and the current in 
the valve ceases to depend upon the value of the anode potential until 
the latter approaches a value of about 20 or 30 volts. 

An important feature of this circuit is that the grid becomes slowly 
negative during the trace and rapidly positive during the fly-back. 
The unavoidable presence of stray capacitance between the cathode 
and earth, including that associated with the deflector plate, has the 
effect of limiting the rate at which the cathode potential can be made 
to fall. If the grid potential is suddenly brought down, the valve is 
completely cut off and the stray capacitance is left to discharge relatively 
slowly through R 2 . By making the sudden change in a positive direc- 
tion, the effect of the stray capacitance is reduced because it tends to 
hold the cathode negative and thus releases a large cathode current 
which charges the stray capacitance. As a result, the cathode potential 
is enabled to change rapidly and the velocity of the fly-back is not 
seriously impaired. 

Another point to be observed is that the valve behaves as a cathode 
follower sufficiently to prevent the output at the cathode from being 
appreciably influenced by the value of the cathode load impedance. 
The presence of R 2i however, increases the effective impedance of the 
valve as seen from the anode, so that the anode current is almost 
independent of the anode potential. The output potential at the anode is 
directly dependent upon the value of R x in parallel with the deflector 
plate leak resistance R 2 and the total stray capacitance associated with 
the anode and the deflector plate. It is for this reason that the stray 
capacitances across # 3 and i? 4 require to be equalized. This balancing 
arrangement prevents the fly-back from overshooting the start of the 
trace and correct adjustment may obviously be obtained by trial and 
error until the early part of the trace has the same velocity as the 
remainder. This may be checked by the application of a calibrating 
oscillator to the Y plates of the cathode ray tube as explained on 
page 143. 

The cathode-follower has a further very important advantage which 
is made use of in this circuit. This is that its input impedance is very 
high so that the leakage across C x is reduced. 

Single -valve Amplifiers employing Negative Voltage Feed-back 

There is a very important class of amplifier which has been derived 
from circuits employing feed-back between the anode and grid of a 
single valve. This class of amplifier has the following characteristics : 









(1) Reduced distortion, 

(2) Circuit operation dependent almost entirely upon the circuit and 
hardly affected by valve parameters except, for example, heater 
and screen potentials. 

(3) Effect of stray capacitance greatly reduced, i.e. very wide fre- 
quency range. 

(4) The circuit can be made to amplify or to attenuate, 

(5) The circuit can be arranged to amplify steady as well as alternating 
potentials. 

(6) The output potential may have its datum at the potential of the 
negative supply rail, i.e. balanced, or it may be either positive or 
negative with respect to this potential, i.e. unbalanced. 

(7) Addition of two or more potentials applied at the grid can be 
obtained at the anode, the output of each being independently 
scaled up or down or left at unity, as desired. 

Of these advantages, the first three are common to all negative 
feed-back amplifiers. 

A Single -valve Balanced Amplifier Employing Voltage Feed-back 

Fig. 138 shows a form of feed-back amplifier* which is capable of 
amplifying steady potentials in addition to alternating potentials up to 
several megacycles per second and 
in which the output has its datum 
at earth potential. In this circuit, 
the resistance /? 3 provides the 
negative feed-back connection. 

In order that the output may 
vary equally on either side of earth 
it is necessary that the potential at 
the point b shall reside at earth 
potential in the "no signal" condi- 
tion. The value of the potentio- 
meter R 7 and of the resistance i? 5 
are chosen to make the cathode 
potential about 2 volts positive with 
respect to earth potential. The re- 
sistances i? 3 and i?,, are made equal, usually about 500,000 ohms. R 1 is 
equal to R 2l the positive H.T. supply is generally about 300 volts and 
the negative bias supply potential is about 150 volts. The potentio- 
meter R 7 is then adjusted until the potential at the point c is 150 volts 
positive with respect to earth. 

# F. C, Williams: "Introduction to Circuit Techniques for Radiolocation/' jfourn. 
I.E.E., 1946, 93, Part IIIA, p, 289. 



OH.T.+ 




-ofarth 



H.T.- 



Fig. 138. Balanced amplifier employing 

voltage feed-back, which amplifies both 

alternating and steady potentials 



200 



TIME BASES 



With the conditions described above and with the input terminals 
short-circuited, the potential of the point b can only be zero provided 
that no current flows in R 2 or R lt since in this case the points a and b 
are at the same potential and a is at earth. Note that no grid current 
will flow because the cathode is 2 volts positive with respect to the grid. 
If, now, the negative H.T. rail is 150 volts negative with respect to the 
earth rail, it follows that the point c must be 150 volts positive with 
respect to earth since R s and R i are equal. The condition of zero 
current in R 2 is brought about by adjustment of the slider of R 7 since 
variation of the screen potential will alter the anode current and, hence, 
the potential of the anode and of the point c. 

Now let the potential of the input terminal be instantaneously 
changed to a potential E IN . Owing to stray capacitance between the 
grid and earth, the grid potential cannot rise instantaneously and there 

will be an initial current % % equal to — ~ flowing in R 1 while the grid 

potential will tend to rise. When this happens, the anode potential 
tends to fall and a current i 2 flowing in the same direction as i x will 
flow in R 2 . This current tends to make the grid potential fall and the 
grid potential will cease to rise when * 2 = *V If an EF50 type of valve is 
used with an anode load of, say, 50,000 ohms, the gain is very high and 
the change of grid potential is extremely small so that the current i 1 is 
substantially unaltered. Therefore, since 



'•> = f = '- 



£>nt!T — 



R< 



'OUT 



<*»--** ~-R t E ">- 



Furthermore, since, in this case, we have made Rj = R z , it emerges 
that E 0UT ~ — E !N . It is thus possible to obtain perfect phase-reversal 
over a very wide frequency range, including steady potentials. Since 
the grid-cathode capacitance of an EF50 valve is approximately 
8 micromicrofarads, the input time constant is not likely to exceed 
10 microseconds and can often be neglected. It will be obvious that, by 
altering the values of R 1 and R 2t a very large range of gain or attenuation 
may be achieved. 

The impedance looking into the output terminals of the valve is 
calculated as follows. 

Let the input terminals be short-circuited and a change of potential 
8e b be applied to the point b, then the change of potential Be s at the 
grid will be ^ R t 



R i + R z 



Se 6 , 



since the potential is applied to the grid via R 1 and R t 



PUSH-PULL DEFLECTION 



201 



Zl 



The resulting change of anode current is g Se g which must come from 
the source which supplies Be b . 

The circuit thus appears to have an impedance due to feed-back 
Se b = R l -\-R 2 1 

g** g R t 'g 

where g is the slope in amperes/volt. 

Since, for a typical pentode, the value of - is about 150 ohms, and 

assuming R 1 =R 2 =i megohm, the output impedance is 300 ohms and, 
obviously, the presence of the anode load resistance i? 6 in parallel with 
it, need not be considered. 

The input impedance of the amplifier may be calculated by assuming 
that a small change of potential Se s is applied to the grid (not to the 
input terminal). As a result, the anode potential will change by — iiBe s . 
The change of potential across R 2 is, thus, Se s —^Be s , and, since these 
potentials are applied to opposite ends of R % > their effect is additive with 
the result that the change of current in the resistance is 

*?> + I) 

and the impedance is — ? = - -. Since u. may be of the order 1,000, 
* |i + I 

this impedance is low. As seen from the input terminals, the value of 

R t must be added because the calculated impedance appears between 

the grid and earth. The impedance of i? 4 may usually be neglected in 

determining the total input impedance. 

Since changes in potential at the input terminal of Fig. 138 cause 

changes at the anode, it will be obvious that changes in the bias potential 

from 150 volts negative will also produce changes, in the ratio — 2 , 

because the circuits are similar. It is thus important to see that the 
negative supply potential is well smoothed. However, an advantage 
exists here in that it is possible to use several input resistances all 
connected together at the grid but having different input potentials 
applied thereto. Each of the applied potentials will appear added at 
the anode and, moreover, by using input resistances of different value, 
the potentials to be added may be arranged to be amplified or attenuated 
by different factors. 

A Single- valve Unbalanced Amplifier Employing Voltage Feed- 
back 

Another form of this circuit in which the output is not balanced 
with respect to earth is shown in Fig. 139. Again the valve is an EF50, 
and the components may have the following suggested values : 



202 



TIME BASES 



OH.T.+ 



Output 




-OH.Z- 

Fig. 139. Amplifier, as in Fig, 138, but 
providing an unbalanced output 



A'. 



R x 1 megohm 

R 2 1 megohm 

R 4 1 megohm 

i? 6 50,000 ohms 
With the H.T. supply rails at 300 
volts positive and 150 volts negative 
with respect to earth potential, ad- 
just R 7 till, with the input ter- 
minals short-circuited, the point b is 
150 volts positive then, if R 2 is equal 
to i? 4 , the potential at the grid 
will be zero with respect to earth, 
provided that the grid takes no 
current. Therefore, the equation 



Eout = ~~^^in n °Jds for this case also, since any change from +150 
volts at the point b will be related to changes of potential from earth at 
the input terminal in the ratio — — 

The output impedance is 

Sfij _ zR x 4- R % _ 1 

gteg ~ R i ~ g 
When i? 4 is made equal to R t and if R t = jR 2 = 1 megohm, the 
numerical value of the output impedance is about 450 ohms. These 
circuits are sometimes called anode-followers because of their resem- 
blance in this respect to cathode-followers. However, the term is a 
misnomer since no following action takes place. 

p 
The input impedance at the grid is which is low but not 

I -\- [A 

usually so low as in the case of the output impedance. As in the case of 
Fig. 138, the value of i? t must be added to obtain the impedance at 
the input terminals. 

If .tf 4 is made large compared with R 2 and if the lower end of 
it is connected to earth instead of to H.T. negative, the anode will 
find its own mean level of steady potential provided a capacitance 
appears between the anode and the grid. Such an arrangement, 
having a further condenser in the grid circuit, is shown in Fig. 140. 
It is not essential for the capacitances to be sufficiently great to pass 
the lowest frequency which it is required to amplify so long as the 
time constants C^ and C 2 R 2 are made equal. 

These devices are often used in electronic computing (see 
Appendix V). 



PUSH-PULL DEFLECTION 



203 



-o Output 



As a further aid to the understanding of these circuits, and to the 
practical design of circuits for modifying the wave-form of an input 
signal in a predetermined manner, the following information is quoted 
from a very valuable paper by Williams, Ritson and Kilburn :* 

The detailed design of wave-form shaping circuits involves the tech- 
nique of synthesizing a cir- ^__ 

cuit to obey a given equation. 
If this is attempted using the 
ordinary methods of rigorous 
circuit analysis, the process is 
laborious in the extreme, and 
some form of approximate 
attack is essential. Since valve 
characteristics cannot be re- 
lied upon to be uniform to 
the required degree of ac- 
curacy from sample to sample, 
negative-feed-back technique 
is essential. The use of negative feed-back also permits the use of 
some very simple approximations which will be established before 
proceeding further. 

The Design of a Feed-back Amplifier Circuit for use in Wave- 
form Shaping Circuits 
Fig. 141 shows a block diagram of the type of amplifier employed. 
The input potential e x is applied to a direct current amplifier of gain 
—A (the minus sign indicates phase reversal in the amplifier) through a 




oh.T.- 



Fig. 140. Balanced amplifier, employing voltage 

feed-back, which amplifies alternating potentials 

from high to extremely low frequencies 



O i / v W- 



a. 



"l 7 



Amplifier' 
(gain = —A) 



-wvw 

R 2 



e 2 
-o 



X 

Fig. 141. Block diagram of a feed-back amplifier 

resistance R v and the output e 2 is fed back to the input through a 

resistance R 2 . This resistance is assumed to be so large that it does not 

appreciably reduce A by loading the amplifier. Now let us suppose that 

the amplifier has been adjusted so that e 2 is zero when the potential e g 

at the input terminal a is zero, and that the gain of the amplifier 

is so high that whenever e 2 departs from zero, the necessary change 

* F. C. Williams, F. J. U. Ritson and T. Kilburn: "Automatic Strobes and Recur- 
rence-Frequency Selectors," Journ. I.E.E., 1946, 93, Part IIIA, p. 1275. 



204 



TIME BASES 



of e^(= — e 2 jA) f is negligibly small. If e x is now instantaneously raised 
from zero to a positive value, then since a is at zero potential, a current 
i x =e x jR x will flow through R x and tend to raise e g towards e x . If e s 
rises, then e 2 falls, and a current i 2 =—e 2 jR 2 flows as indicated in the 
diagram, tending to draw off the current i from the amplifier input 
terminal a. If the input resistance of the amplifier is infinitely large, 
then any excess of i t over i 2 will still tend to raise e g and so depress e 2 
still further until ultimately the whole of i x will be drawn off by i 2 . 
The conditions in the circuit are now such that 



U = t t 



and 



*i/*i - ~ e t iRi 



or # «2= -(* 2 /*i>i- 

The time taken for the circuit to reach this condition depends on the 
response time of the amplifier. In our problem this will be so short 
that the process can be regarded as instantaneous. It is, of course, 
essential to the proper operation of the circuit that the amplifier response 
be such that it is stable when fed back. The requirements for stability 
have been dealt with extensively elsewhere* and need not concern us 
further here. 

The Concept of the Virtual Earth 

In the process described above, the function of the amplifier is to 
hold the potential of a very near to zero in spite of the change in e x by 
drawing off i x through R 2 by an appropriate change of e 2 . Thus, if we 
assume that a virtual earth exists at a and that the current i x flows on 
past this virtual earth through R 2 to generate e.,, we can omit the 
amplifier from our calculations entirely. This is indicated in Fig. 142. 



1, = e,/R, 



O ^AAAA' 



-AAAA- 



1 * 2 *.&* 



o— I Virtual earth . — o 

Fig. 142. Schematic diagram of a feed-back amplifier showing a virtual earth 



Although we have described the mechanism entirely in terms of resis- 
tances and steady potentials, it is also valid when R x and R 2 are replaced 
by Z 1 and Z 2 , provided that Z x and Z 2 are not such that stability is 
impaired. In addition, it holds for alternating potentials and transients, 
provided that all frequencies concerned are low compared with the 
amplifier cut-off frequency. Z x and Z 2 are now to be treated as operators 

• H. W. Bode: Network Analysis and Feed-back Amplifier Design, D. Van Nostrand 
Co. Inc., 1945. 






PUSH-PULL DEFLECTION 



205 



in the manner described on page 207 et seq. Thus, as a general case of 
Fig. 142, we can draw Fig. 143. From which, 

i 1 =eJZ 1 •••W 

e 2 = — Z 2 i x . . . • (2) 

Hence e 2 m — (Z 2 fZ x )e x . . . . (3) 

It will be adopted as a principle of design that, wherever possible, 
conditions will be chosen so that the above simplification is per- 
missible. To assist in checking that this has, in fact, been done in a 
particular case, it is useful to have formulae for the errors involved in 
equation (3). 



Q 



1 



*1 



Virtual earth 



^7 



XT' 



Fig. 143. General case of the feed-back amplifier 

The Estimation of the Errors introduced by using Equation (3) 

The source of error lies in the fact that the potential of the virtual 
earth is not in fact zero, but 

e g = -ejA. . ..(4) 

Hence the current in Z x is not i x — e x !Z l but 

i\ = (e x - e g )jZ v 
The input current has been changed by a fractional amount —eje l 
and, as the output potential is proportional to i\, it will be changed by 
a similar fraction. If this fractional change in output potential is written 
as Sj, then 

h - - eje x . ... (5) 

The above equation can be re-expressed, using equation (4), as 

8 X = + e 2 (Ae x 
and using the approximate expression for e 2 given in equation (3) it then 
becomes 

*, = - (i/A)(Z 2 !Z x ). ... (6) 

The second effect of the departure of e g from zero is that a given current 
1 flowing in Z 2 now produces an output potential 

e' 2 = — Z 2 i + e r 
The output voltage has therefore been changed by a further fractional 
amount S 2 where 

8 2 = eje 2 = - ijA. ... (7) 



2o6 



TIME BASES 



A third source of error arises in practical cases where a third im- 
pedance Z 3 exists between the amplifier input terminal and earth. The 
potential e g will appear across this impedance and the current through 
it will be lost from Z 2 , hence e 2 will change by a further amount 
{ e jZ^)Z 2 . This may be expressed as a fraction of e 2 and, since e g = — e 2 (A 



8 3 = £ 2 fi 



±11. 

Z*A 



(8) 



It should be noted from equations (6), (7) and (8) that all the errors 
are inversely proportional to A so that by making A sufficiently large 
they can be brought within any desired limits. 

The error S 2 can be dismissed entirely by noting that it is less than 
i per cent provided A > 100, but the errors S 2 and S 3 can only be dealt 
with for each particular case as it arises, since Z v Z 2 and Z 3 are usually 
dissimilar impedances. 

The Output Impedance of Feed-back Amplifiers 

Since, in general, we shall require to follow the transformation of 
e x into e 2 by a further transformation which will involve loading e 2 to 
some extent, it is important to know how such loading will affect the 
value of e z , First, viewing the matter qualitatively, loading of the output 
terminal is equivalent to a reduction of A, so we may say at once that 
any load on e 2 which does not seriously affect A will have a negligible 
effect on e 2y since by hypothesis A is so large that its value does not 
enter into the approximate evaluation of e 2 . 

This is equivalent to stating that the feed-back amplifier has a low 
effective output impedance. Mathematical evaluation of this effective 
impedance in terms of Z x and Z 2 is of little value except when referred 
to specific values of Z x and Z 2 and specific input wave-forms. In 
general, the output impedance is small, being of the order of 500 ohms 
in a typical case (see page 202). This may usually be neglected in com- 
parison with the impedance that will be connected to e 2 , which will 
normally be of the order of 100,000 ohms or more. We are therefore 
entitled to connect a second amplifier directly on to the output of the 
first, and there is no reason why further amplifiers should not be con- 
nected in cascade if required. 

If the impedances in the input and output circuit of the second 
amplifier are Z 3 and Z 4 respectively, and if c 3 is its output voltage, then 
it is apparent by using the method already described, that 



H = {ZjZ z )e % = (ZjZ^ZjZJe,. 



(9) 



Sometimes the input may exist as a current, rather than as a potential. 
In these cases Z x is omitted and the input current i % replaces e x (see 
Fig. 144). 









PUSH-PULL DEFLECTION 



207 



The Input Impedance of Feed-back Amplifiers 

Since, in fact, no current source is perfect, i.e. of infinite internal 
impedance, it is of importance in these cases to know what impedance 
is presented to i x by the feed-back amplifier. 

This is easily shown to be 

Zi = Z t l(A + 1) • • ■ (10) 

either by comparison with the well-known Miller effect or by using 
the methods described on page 201. 






i 



Virtual earth 



e 2 = -2 2 t, 
o 



o -J- 

Fig. r44. Feed-back amplifier with current input 

It follows that any earlier stage works into an impedance similar in 
physical form to the feed-back impedance, but whose components are 
all such as will present i((A-\-i) of the impedance: this is true for all 
frequencies or frequency components for which A is not complex. 

For example, if Z % comprises C 2 in series with R 2 , the input impe- 
dance is (A -\-i)C 2 in series with RJ(A -f 1). If A is large it is permis- 
sible to assume that the earlier stage works into a short circuit. 

The Shaping Effect of Typical Amplifier Circuits 

Table IV shows a set of typical input and output networks together 
with the shaping effect on a step function of either amplitude or rate 
of change of amplitude. The notation and method of using the 
operators is described on page 332, where the method of splitting 
up the operator and applying the component operator step by 
step is also explained. The admittance of the input arm and the 
impedance of the output arm are expressed as operators under the 
relevant arms: the total operator is the product of these quantities. 
Where current feed is used in place of potential feed, any of the circuits 
shown in column 1 may be used with the input current i replacing 
e x jR v There are, of course, innumerable other networks which may 
be used both for input and output arms : in particular there are a great 
many three-terminal networks in which one or more of the elements 
has one terminal earthed. Two useful examples are given in Figs. 
145 and 146. In Fig. 145, i x is calculated on the assumption 
that the virtual earth is at earth potential and the output potential is the 
product R 2 i v In Fig. 146, the process is to calculate i 2 in terms of e 2 , 



208 



TIME BASES 



Table 



R> 



Tl 



# 



-w#- 



/> ^ W\A— 1 ■ -'VW- ■ o 

^*2_ Vfrttajearth r e ^ 

* Admittance Impedance „ ^ 

**£&> -LI..** 



«-f "^i/-| 



-#* i ^ £■ DC =L 



f 
HH^ 



Y K 



e,= 



_ / 



Slope ■ 



<?/! 






*_. — -II 1 — 



e s 



pU%%) 



CR % 



erftii fin 



%*% fin 

fy Rz c> 

1 4- r 



pQRf c O) 



°2 n 



i c, $n 



JLK* 



Q ^ 






c, 
AV 



-wv 1 



/-< — $iope=(±Q 



**4 "Qi* 



*- w< 



A 



"1 



T 



LfcTr 






% J! 



ft 



-Ez I - /S-R z q I-&&, 



13 






% 



A 






i-i 






PUSH-PULL DEFLECTION 



209 



IV 



C„ R, 



~U> 









X " r 



%e 2 



t+pCRj 



7 r, 



ft* — r — I 






A ii^ 



-I II— T*ft 



1 c 



er 



'/ p 






R t 



e-5 



Z~-R, 



r 






t'c/fjvwjfy r=c > R > 



% •: *M w> 1 * £ M a 



ft 



'/7^q 






''/-v — r 









K=£^ 



//pf/^, 



r^fir 






15 






"I "^ 






' ; 1 { ^4^ I X^ 



y^jwAf z JJ p^ 

.Mtsm (ties® f&p^cMM 

4rp R, pQJL^n' 2 « 



12 



A L* — ^ 

' Y* 1+PbRi Z '/^R 2 
R 2 (1+pCA) f±Q Ir, % 



16 



210 



TIME BASES 



and equate it to i x . If the operators, of the form eji ty are defined as 
transfer operators, the general statement can be made that, with potential 
feed, the output is the transfer operator of the output circuit divided 
by the transfer operator of the input circuit. With current feed, the 
output is the input current at the virtual earth multiplied by the output 
transfer operator, 



i 



;prr 



I 



Virtual earth 

— % — 

Fig, 145 



' 1 ■:- pCRj 



V\A/ — O 
*2 



X^T 1 -* 



I 



Virtual earth 




Fig. 146 

i 



1 +pCR 1 

1 +P(C l + C 2 )R t 
1 + P(Cj + CJRt 



2 ptCiCMi+pCRJ ' 
Examples of feed-back amplifier circuit using three-terminal networks 

A list of networks and their transfer operators is given in Table V, 
which is taken from a paper by Williams and Uttley* and this table may 
be used when it is required to design a circuit to obey certain equations. 
The operators are used as they stand if the network is in the output 
circuit and inverted if it is in the input circuit. For the process of 
synthesis this table largely supersedes Table IV. 

(c) Two-valve Push-pull Amplifiers 
Standard Forms of Push-pull Amplifier 

Although it is possible that one might desire to amplify the output 
of a time base which itself provides push-pull potentials, the more 
usual requirement is to provide an amplified push-pull output from 
an unbalanced input, f The ordinary push-pull amplifier of the type 
which requires a push-pull input is discussed later in this chapter. 

Fig. 147 illustrates a method J which was often employed before the 
cathode-coupled push-pull amplifier was developed. 

The ratio of the resistances i?j and R 2 is chosen so that the gain 
from the anode of V 1 to the anode of V 2 is approximately unity and 
the operating conditions of the valves are carefully chosen to allow 

•F. C. Williams and A. M. Uttley: "The Velodyne," Jottm. I.E.E., 1946, 93, 
Part IIIA, p. 1256. 

t W. T. Cocking: "Push-Pull Input Circuits," Wireless World, 1948, pp. 7, 62, 85, 
126 and 183. 

t R. E. H. Carpenter: British Patent 325833 ; O. S. Puckle: British Patent 432268. 



PUSH-PULL DEFLECTION 






'<■!■ 



B 
PQ 




K 



ttj 







II 



:t 



jr 



<u 



CN 




g 




«!■* £ 



211 



2 j= 



C 



§ I 



to £ 



212 



TIME BASES 



the curvature of the characteristic of V t to compensate for the curvature 
of V v Thus, the output potentials relative to earth are both slightly 
distorted while the potential difference between the output terminals 
is comparatively un distorted. This method is difficult to apply in 
practice because the adjustments necessary to reduce distortion to a 
minimum are rather critical and laborious. 

If pentode valves are employed, the screens may be fed from a 
common series resistance connected to the H.T, supply rail. Only a 



OffX* 




OH.T- 



Fig. 147. An amplifier providing a push-pull output which is 
obtained by means of a phase-reversing stage 

small decoupling condenser is generally necessary if the valves are not 
overdriven since the screen currents are 180 out of phase and the 
resultant potential is constant. If desired, the cathodes of the two 
valves, whether triodes or pentodes, may employ a common bias 
resistance and decoupling condenser. 

In general, the cathode-coupled form of amplifier will prove much 
more satisfactory for time base amplification. 

Cathode-coupled Push-pull Amplifiers 

When the output potential of a time base with the addition of a 
phase-reversing stage is insufficient to provide the required push-pull 
deflection potentials, a cathode-coupled amplifier will generally provide 
a satisfactory solution. This device is suitable for amplifying either 



PUSH-PULL DEFLECTION 



213 



the horizontal or the vertical deflection potentials and its attractive 
features include; 

(1) Low distortion and relatively small value of grid current when 
overloaded. 

(2) Freedom from any tendency to self-oscillation. 

(3) Absence of "jitter" or hum derived from fluctuations of the 
H.T. supply potential. 

(4) Facilities for arranging D.C. connections at the input and output. 
($) Facilities for providing shift, astigmatism correction and balance 

controls. 
(6) Facilities for providing sweep expansion. 

In the case of oscillographic investigations at very low frequencies 
the cathode-coupled amplifier is extremely popular and is almost 
invariably employed as the output stage of the signal amplifier. For 
time bases, particularly if the repetition frequency is high, the utility 
of such an amplifier is limited because it is difficult to obtain a rapid 
fly-back when the anode load resistances are sufficiently high to mini- 
mize the effects of curvature of the valve characteristics. However, 
the features noted above make it very suitable for providing an expanded 
sweep. 

The amplifier is preferably directly coupled to the deflector plates, 
so that the spot lies at the centre of the screen when the two anode 
potentials are equal. This automatically provides a foolproof means 
of adjusting the relative bias on the valves, because the bias adjustment 
serves as the shift control, and if the valves are wrongly biased with 
respect to each other the image will be deflected to one side of the screen. 

Cathode -coupled Amplifiers 

The circuit of a cathode-coupled push-pull amplifier is shown in 
Fig. 148. In this circuit, which is basically due to Blumlein,* the 
potential across the time base condenser is applied between the grid of 
V x and the grid of V 2 , which is held at a fixed potential E 2 . 

Provided that the common cathode resistance R 3 is made large, so 
that the valve V x is capable of behaving like a cathode follower when 
the other valve is removed or its anode current cut off, the sum of 
the anode currents in the two valves will be nearly constant when both 
are conducting. Thus, any reduction in anode current of one valve is 
almost exactly balanced by an increase in that of the other. Normally, 
R x and R 2 are made equal and two similar valves are used, but it is 
not usually found necessary to obtain specially matched valves. In 
order to secure the best results, the values of the resistances should be 

* A. D. Blumlein: British Patent 482740. 



214 



TIME BASES 



chosen in relation to the values of the fixed potentials E x and E 2 and 
the valve characteristics. 

When the input terminals are short-circuited, the input potential is 
zero and the valves are equally biased by the difference between the 
potential drop across i? 3 and the fixed potential E 2 . The latter is nor- 
mally made ten or twenty times the grid base (i.e. the grid volts required 
to cut off anode current when the potential between the anode and the 
cathode is E x —E 2 ) so that a small percentage increase in the combined 
cathode currents is sufficient to change the grid-cathode bias potential 
from zero to the cut-off point, or vice versa. Under these conditions the 



0%+ 




Input 
O — 



f* 



-oE 7 + 



\R, 



-O/f.T.- 



Fig. 148. A cathode-coupled push-pull amplifier 

total cathode current is automatically regulated at a value approximately 

E 
equal to -— and is not greatly dependent on the valve characteristics or 
R 3 

the values of R x and R 2 , If one valve is removed from the amplifier the 
potential across i? 3 becomes slightly less and the grid-cathode bias on 
the remaining valve falls, allowing its anode current to increase so that 
the current through i? 3 remains nearly at the original value. 

When the grid potentials are unequal, as is the case when an input 
potential is applied to the grid of V x , the total cathode current is 
roughly proportional to the average of the grid potentials, measured 
from the negative rail, so long as neither valve is cut off. Thus, if the 
grid potential of V x is raised, say 4 per cent, above E 2 , the average 
grid potential rises by 2 per cent and the total current increases by 
about 2 per cent although V x may now take practically all the current. 
The increase of current in V 1 is obtained almost entirely at the expense 
of the current in V 2 . The anode currents of the two valves are com- 
plementary, and if the anode loads are balanced almost equal push-pull 



T 



■ 






PUSH-PULL DEFLECTION 



21 



potential changes are developed. Exact equality of the output potentials 
from the two anodes is obtainable at low frequencies by making 



R 2 = 



/tf,fo, + 1) + R a2 \ 



where \l 2 and R aZ are the amplification factor and anode impedance 
respectively of V 2 . If it is desired to produce an output having some 
ratio of potentials other than 1:1, it is only necessary to adjust the 
anode load resistance values to the desired ratio. 

When the input potential is sufficiently negative, V x is cut off and 
V 2 takes its maximum current, although grid current does not flow. 
Beyond this point, the input potential may be made still more negative 
without producing any further change in either valve. On the other 
hand, if the grid of V 1 is made sufficiently positive to cut off F 2 , then 
the cathodes begin to follow the grid of V x and the bias on this valve 
only changes slowly with further increase of grid potential. Although 
this change is slow by comparison with the changes which occur when 
both valves are conducting, a point is reached at which the grid-cathode 
bias is low enough for grid current to commence and further increase 
of grid potential will produce an increase of grid current. The value 
of the grid current corresponding to a given overload is small when 
compared with that in an ordinary amplifier without the protection 
provided by the high cathode load resistance. 

Since the changes of anode current in the two valves are nearly 
equal and opposite, the ratio of the output potentials at the anodes is 
substantially the same as the ratio of R x to R z> even if the anode loads 
are very unequal. One aspect of this condition, due to the presence 
of jR 3 , is that the output impedance at each anode is high and therefore 
the output potentials relative to earth are apt to be disturbed by the 
connection of external circuits which may apply unequal capacitances 
or resistances in parallel with J?! and R 2 . When the external impedances 
have the same ratio as the anode load resistances or appear only between 
the anodes, instead of between anode and earth, the effects are no 
worse than would be obtained if the same valves and effective anode 
loads were employed in a single-valve amplifier or a push-pull amplifier 
of ordinary type, e.g. as shown in Fig. 147. 

In order to reduce amplitude distortion, the anode load resistances 
should be made large in comparison with the internal anode impedances 
of the valves and the two valves should be of the same type. The high 
impedance of pentodes makes their use unprofitable in this circuit 
because the anode load resistances cannot be made high enough to 
reduce distortion. Pentodes and tetrodes may, of course, be used if 
their screens are connected to the anodes so that they behave as triodes. 
The reason for the comparative absence of amplitude distortion may 



2l6 



TIME BASES 



be seen from the equation given below. If the amplification factors of 

V-i and V 2 are equal and the fixed potential E 2 is large compared with 

the grid base, the gain A x from the grid of V 1 to the anode of V x is given 

approximately by n 

£ __ I**m f 

^al+Ka+^l + ^2 

where jj. is the amplification factor, and R al and R aZ the anode im- 
pedances of V x and V 2 respectively. 

In a triode, \i is relatively constant while R a varies considerably, 
depending on the anode current. In a cathode-coupled amplifier, with 
a pair of similar valves, the variations of R a tend to cancel each other 
because R al increases when R a2 decreases and the gain depends only 
on their sum. The changes of (/? a i+i? a2 ) are rendered ineffective by 
making (i? 1 +i? 2 ) relatively large. If R 2 is made equal to R t the gain 
becomes, approximately / 7? \ 

where R a is the nominal anode impedance of one valve. This is half 
the gain which would be anticipated in a straightforward single- valve 
amplifier. If the resistance i? a is split into two parallel sections, one in 
each cathode lead, and if a variable resistance is placed between the two 
cathodes which are, now, not directly connected together, the variable 
resistance will serve to control the gain of the amplifier.* 

When negative current feed-back is obtained by the addition of an 
unshunted cathode resistance which is common to both valves of a 
push-pull amplifier, a part of the input potential appears between 
cathode and grid and part between the cathode and the negative 
supply rail so that the effective grid base, as seen from the input 
terminals, is extended. This circuit has been described as a "long- tailed 
pair." When the feed-back is removed by the insertion of a gain control 
resistance between the two cathodes, the circuit behaves less as a 
long-tailed pair as the resistance is increased in value. 

In order to design a cathode-coupled amplifier, one may choose the 
value of the anode load resistance R t in the ordinary way, making 
it large compared with the value of R ai while the required H.T. potential 
E x may be determined by allowing E 2 volts above the value which 
would normally be required. E 2 should be as high as can conveniently 
be obtained, at least ten times the grid base corresponding to an anode 
potential equal to E x —E 2 . The required value of R% is then determined 
from the equation (^EJfft^Rj 

/<3 Ei-v-E.-E^ 
where E is the grid base of the valve for an anode potential of E t ~E Z . 
*M. L. Jofeh: British Patent 529044. 









PUSH-PULL DEFLECTION 

The above equation states the 
conditions necessary to avoid 
grid current in V x before V 2 is 
cut off. Fig. 149 depicts the 
conditions existing when V 2 is 
at the cut-off point and it will 
be seen that the grid-cathode 
potential of V x has been fixed E 2 +'E -i 
at minus one volt in order to 
prevent the flow of grid current. 



217 



o£ M 






(■qRq, "MS qc = i a %a "A (-*) 



*3° 



ia.R3=E 2 +E 



O/f.T- 



If R 3 is made variable, the 



Fig. 149. 
cathode 



The conditions existing in the 
■coupled amplifier of Fig. 14S 



minimum value should not be 

less than that given by the 

above equation. The value of R 2 may be determined from the equation 

given on page 215, or, if precise symmetry is not demanded, R 2 may be 

made equal to jt la 

It is to be noted that a cathode-coupled amplifier may be designed 
with comparatively small values for E 2 and i? 3 and yet the output 
potentials may be kept balanced by making #2 considerably larger than 
R v However, in a design of this type, the variations of anode im- 
pedance in the two valves do not compensate each other and distortion 
is likely to be greater than is the case when E 2 and R 3 are large. 



•OH.T.+ 




OH.T.~ 



Fig. 150. Clare's push-pull amplifier 



Clare* has suggested a further modification to this circuit which 
provides push-pull amplification even at extremely low frequencies. 
The circuit arrangement is shown in Fig. 150, from which it will be 
noted that the grid of V 2 is connected directly to the anode of the 
previous unbalanced valve stage V x , and the integrator circuit /? 6 C 2 
ensures that the bias potential at the grid of V z is the same as that at 

* J. D, Clare: "The Twin Triode Phase- Splitting Amplifier," Electronic Engineering, 
February 1947, p. 62; R. L. Gordon: "A Note on the Twin Triode Phase-Splitter," 
Electronic Engineering, August 1947, p. 16. 



2lS 



TIME BASES 



the grid of V 2 in the absence of a signal. When a signal is applied to 
the grid of F 2 , via the valve V lf provided the time constant of R 6 C 2 is 
sufficiently large it is not passed to the grid of F 3 , which therefore 
remains at constant potential. On the other hand, if the potential of 
the H.T. supply rail, or of that at the anode of the valve V lt alters for 
any reason, the bias at the grid of V 3 is automatically adjusted to 
equality with that at the grid of V 2 so that the balance of the output 
potentials at the anodes of V 2 and V 3 remain substantially unaltered. 

The Amplification of Balanced Potentials 

The cathode-coupled two-valve amplifier circuit may be arranged 
as shown in Fig. 151 to provide amplification of two potentials which 



OH.T + 




Fig. 151. 



J — O H. T. - 



A cathode-coupled amplifier, having a balanced output and which 
discriminates against in-phase input potentials 



are balanced with respect to earth and, at the same time, to discriminate 
against unbalanced potentials (e.g. hum) appearing in phase at the two 
input grids. 

If an input potential which is balanced with respect to earth is 
applied to the two grids in push-pull, the potential of the two cathodes 
remains sensibly constant while the valves are operated over the linear 
part of the characteristic. In this case, an increase in V y anode current 
is accompanied by an equal decrease of the anode current in V z . The 
potential across R 4 therefore remains unchanged and the valve V r does 
not behave as a cathode follower. As soon as equality of input of the 
potentials applied to the two grids ceases to obtain, the cathodes com- 
mence to follow the grid to which the greater amplitude of input is 
applied. The gain of that valve is reduced until equality of output 
potential is once more achieved. If the phase difference between the 
two inputs ceases to be 180° the cathodes follow first one grid and then 






PUSH-PULL DEFLECTION 



219 



the other, thus reducing the gain of both valves and reducing the output. 
For in-phase inputs, the gain is greatly reduced. 

The circuit of Fig. 151 and various modifications thereof are due to 
Blumlein,* who was the first to point out that, if the cathode resistance 
R^ is sufficiently high in value to cause each valve to behave as a cathode 
follower in the absence of the other, no feed-back occurs for balanced 
input potentials over the linear portion of the valve characteristic and 
a large degree of feed-back occurs for in-phase potentials. This property 
of the circuit is of great value, particularly in those cases where the 
source of the input potential is of high impedance since, in these 
circumstances, the electrostatic pick-up of hum potential is a maximum. 
Obviously, the effect of electromagnetic pick-up in wiring loops is 
also reduced. 

This type of circuit is not a long-tailed pair (see page 216) since, for a 
balanced input, neither valve behaves as a cathode follower and hence 
there is no feed-back. 

The total gain is approximately twice that obtained when one grid is 
held at a fixed potential as is the case with a long-tailed pair. 

If this circuit is used to provide push-pull deflection of a cathode 
ray tube spot, the hum appearing at the grids will not deflect the spot, 
but it can cause focusing troubles. An in-phase input to the grids will 
cause the mean level of the steady potential at the two anodes to vary 
and, as explained on pages 276 and 278, this can lead to deflection 
defocusing and to astigmatism. 

As already explained, the greater is the value of i? 4 the more effective 
are the balancing and in-phase discrimination properties of the amplifier. 
For this reason R 4 may be replaced by a pentode valve operated above 
the knee of the curve. It is usually necessary to make suitable bias 
arrangements for the cathode-coupled amplifier by connecting its grid- 
leak resistances to some point which is positive with respect to the 
cathode of the cathode-load pentode which replaces J? 4 . 

Cathode-coupled Amplifier Stages Connected in Series 

Mansfordf has developed an arrangement of cathode-coupled pairs 
connected in series across the H.T. supply rails. This system is satis- 
factory only because the total current of each pair remains constant 
while the valves are operated over the more linear portion of the 
characteristic. This is, of course, a necessary requirement, since other- 
wise a reduction of current in one stage would cause a similar change 
in all the remaining stages with a consequent large reduction in gain. 

* A, D. Blumlein ; British Patent 482740. 

t H. L. Mansford: British Patent 579685; D. L. Johnston: " Electro-encephalograph 
Amplifier," Wireless Engineer, 1947, 24, pp. 231, 277 and 292. 



220 



TIME BASES 



Fig. 152 shows the basic arrangement of the circuit- The valve V 1 
may, of course, be replaced by a resistance or two separate cathode 
resistances or pentodes may be employed and a variable resistance may 
connect the cathodes of V 2 and F 3 for providing variable gain control. 
If V 1 is employed, the resistance R 7 should not be less than 5,000 ohms 
and should preferably be 10,000 to 20,000 ohms. The resistance R x2 



Out 




<3H,T.+ 






In 
Fig. 152. Cathode-coupled amplifier having stages connected in series 

is provided to enable the correct value of bias potential to be applied 
to the valves F 2 and F 3 . The screen of V 1 may be connected to earth 
if the potential at this point is suitable. Special biasing arrangements 
may be necessary for the valves F 4 and F 5 . It is essential to tap the 
grid from the anode end of i? 4 and i? 6 , so far as the signal is concerned, 
since, otherwise, the maximum gain available is not realized. If steady 
potentials are required to be amplified, the bias may be increased by 
the insertion of resistances, or a single resistance, between the cathodes 
and the junctions of i? 4 , i? 5 and R 6 . The bias may be reduced by 
inserting resistances of appropriate value between the anode and grid 












PUSH-PULL DEFLECTION 



221 



and between the grid and a source of constant potential as shown dotted 
in Fig. 152. The resistance R a is employed to enable valves having a 
larger steady anode current to be employed for V \ and F 5 than is 
required in the case of F 2 and F 3 . The difference between the sum of 
the anode and screen currents for the valves F 4 and F 5 and that for 
F 2 and F 3 flows through R B . Obviously, F 2 must carry the combined 
anode and screen currents of F 2 and F 3 . Extra stages may be added 
in the same manner. 

The use of a pentode valve, operated above the knee of the curve, 
has the advantage that the value of the common cathode resistance can 
be raised to something of the order of 1 megohm without an inordi- 
nately large potential drop across it. The effective anode to earth 
impedance of a pentode so used is i?„+i? 7 (i +ji), where R a is the normal 
anode impedance of the valve and (i is the amplification factor. 

Circuits arranged in the manner described have the following 
advantages: 

(1) The H.T. supply for the cathode ray tube may be employed for 
the amplifier. 

(2) No decoupling condensers are required. 

(3) Ripple on the positive H.T. supply rail is in phase on the two 
output terminals, but it cannot be introduced at an early stage, 
nor can it be amplified in the chain. The effect is, therefore, 
generally negligible. 

(4) Because of the discrimination against in-phase components, 
instability due to stray-capacitance feed-back is greatly reduced 
and is almost an impossibility. 

A. Cathode -coupled Amplifier Time Base Providing Facilities 
for Sweep Expansion 

Fig. 153 shows a circuit for a time base in which a cathode-coupled 
amplifier is arranged to provide sweep expansion. The condenser C x 
is rapidly charged from the H.T. supply rail by means of any suitable 
trigger device or gas-filled relay tube (indicated by the key K) and 
discharged slowly through the constant- current device D. The potential 
across C\ is applied to the anode of the diode F x and the grid of the 
triode F 2 . The effect of the diode is to bring the grid potential of F a 
to a definite potential each time C x becomes fully charged. Any 
leakage of the charge on C 2 is also compensated. The diode therefore 
acts in the same way as a D.C. restorer and fixes the potential at the 
commencement of the trace without reference to the extinction potential 
of the switching device K. The valve F 2 is a cathode follower and 
the potentiometer R z and resistance i? 4 are effectively in parallel since, 
from an A.C. point of view, all the supply rails are at the same potential. 



222 



TIME BASES 



The change of potential applied to the grid of V 3 is thus adjusted by 
altering the position of the slider of i? 3 , but the value of the resistance 
R 2 is so chosen that the potential at the grid of V 3 at the commencement 
of the trace remains constant, irrespective of the adjustment of the gain 
potentiometer R z . Thus, when the key is closed, R 2 is adjusted to 
make the lower end of R 3 have the same potential as its upper end. 

O+/60V 




OOV 
O-I50V 



-325 V 
'XSh/ft 
R-3 Co/ftro/ 

O-340V 



Q-440V 



Fig. 153. A cathode-coupled amplifier time base providing facilities 
for sweep expansion 

The trace on the cathode ray tube screen therefore always commences 
from a fixed point. 

Since the potential change across C x is far more than is necessary 
to cut off V % and F 3 the output potential on the opiates may be 
expanded until the whole of the trace is produced by about 25 volts 
out of possibly several hundred available across C lt the remainder 
producing no further deflection. As the potential applied to the grid 
of V z is reduced, a greater percentage of it produces the trace and, as a 
result, the velocity of the cathode ray spot is reduced. Thus the sweep 
expansion does not increase the length of the trace, but it increases 
the spot velocity so as to open out the first part of the trace. 

The value of the fixed bias on the grid of F 4 is controlled by adjust- 
ment of R 9 . This has the effect of moving the trace bodily in the X 
direction. Adjustment of the resistance R 7 controls the degree of 









PUSH-PULL DEFLECTION 



223 



astigmatism because it changes the mean potential on the anodes of the 
valves with respect to the final anode of the cathode ray tube (see 
Appendix I, page 278). 

The percentage of the repetition period represented by the trace is 
normally controlled by the expansion potentiometer R z but, if it is 
desired to raise the percentage to too, it will be necessary to adjust the 
trigger device K so that the maximum potential change across C x is 
only about 50-60 volts. 

The following values may be used for the component parts: 



Rv 


2 megohms 


R* 


500 ohms (variable) 


R 3 and R 4 


20,000 ohms each 


R b and R & 


100,000 ohms each 


R, 


10,000 ohms 


R* 


30,000 ohms 


*a 


50,000 ohms 


i? 10 and J? u 


2 megohms each 


C x and C 2 


dependent upon repetition frequency 


c 3 


not less than 1 txF. 


C 4 and C 5 


dependent upon the frequency of the signal under 




examination 


V% 


Di, EA50 or DDL4 



F 2 ,F 3 andJ/ 4 MH 4 . 

Since the cathodes of the valves, except that represented by D, are 
all at approximately the same potential, their heaters may be fed from 
a single winding on the transformer. The heaters should be connected 
to the 325 volt negative supply rail. The constant-current valve D 
should have its heater fed from a separate winding connected to the 
440 volt negative rail. 

The time constant C 2 Ri must be large compared with 7\ the repeti- 
tion period of the time base, in order not to introduce curvature of 
the wave-form. C 3 is made large to eliminate hum on the grid of K 4 . 
The output obtainable from the time base using the components 
specified is of the order of 300 volts. 

Anode -coupled Push-pull Amplifiers 

Carpenter* has described an anode-coupled push-pull amplifier, the 

circuit of which is shown in Fig. 154. In this circuit, the potential which 

drives the grid of the valve V 2 is obtained from the junction of a pair 

of equal or nearly equal resistances connected between the anodes of 

* R. E. H. Carpenter: British Patent 325833; "See-saw or Paraphase?", Wireless 
World, 1945, 51, p. 263; M. G. Scroggie: "The See-saw Circuit," Wireless World, 
1945, 51, P- 194- 



224 



TIME BASES 



V x and F 2 . If the changes of anode potential on each of the valves is 
the same and if the two resistances R 3 and R 4 are equal, the potential 
change at the slider would be zero, no change of potential would appear 
at the grid of V 2 and, hence, there would be no change of potential at 
its anode. Manifestly, therefore, there must be a difference of potential 
swing on the two anodes or, alternatively, the resistances R % and i? 4 
must be unequal. 

The term "see-saw" has been applied to this circuit because, as one 
end of i? 3 rises in potential, that of R 4 falls, their junction remaining 
sensibly, but not quite, at the mean potential which the two anodes 




OHX+ 



Output 



X 3 



% 



MW WW WW-^ 

t 



n 



Input 



^T C! 




FlG. 154. Carpenter's anode-coupled push-pull amplifier 

assume when no signal is applied and the input terminals are short- 
circuited. 

It will be obvious that the higher is the gain of the amplifier valve 
V % > the more nearly equal can the two resistances R B and i? 4 be made. 
It will also be obvious that, if it is desired to have unequal outputs 
from the two anodes, all that is necessary is to alter the ratio R s jR 4 . If 
one of the valves tends to lose emission, the degree of balance is 
retained at the expense of reduced gain in the other valve, i.e. at the 
expense of reduced overall gain. Since the grid of V 2 must be driven 
by a potential which is in the same phase as that of the anode of V lt 
the junction of the two resistances from which the signal is applied to 
the grid must reside at a point in the i? 3 -K 4 chain which is to the left 
of the zero potential point. Thus, the value of the resistance R a must 
always be smaller than that of i? 4 unless it is required to have a smaller 
output from V 2 than that given by V v 

In order to adjust the balance of the circuit, one of the following 
two procedures should be adopted: 

(1) Connect a low value of resistance in the common H.T. supply 
rail to the two valves and examine the alternating component of 
the potential which appears across it with an oscillograph having 
an input amplifier. The potentiometer forming part of the 






PUSH-PULL DEFLECTION 



22 = 



resistances R. d and R A should then be varied until the potential 
falls to zero with the oscillograph amplifier gain set to maximum. 
The potentials appearing at the two anodes will then be in strict 
proportion to the values of the anode load resistances and will 
not necessarily be equal. 
(2) Connect the output terminals to the Y ± and F 2 terminals of a 
single beam oscillograph via two amplifiers. The sensitivity of 
the cathode ray tube and amplifiers should first be matched by 
applying a fixed amplitude of alternating potential to one amplifier 
while the other is short-circuited, and noting the deflection. Then 
carry out the same procedure with the second amplifier connected 
to the other Y plate of the oscillograph and adjust the gain till the 
deflection is the same as before. When the output from the 
see-saw circuit is connected to the amplifiers, adjust the slider 
till no signal appears on the cathode ray tube. In these conditions, 
the output from the two valves of the see-saw circuit will be equal 
for the particular values of the anode loads resistances employed. 



+ 30% 




-40% 



-30% -20% -70% +10% +20% +30% 



*W9 



Fie. 155. Relation between the output unbalance S- and the resistance unbalance 

ZR for different values of gain A of the V 2 amplifier stage of Fig. 154, when Ry 

and i? 2 are of equal value, J? ftl equals R*t 2 and R$ is equal to £ (iJg+jRj) 

Fig. 155 shows the relation between the anode potential unbalance 



Se - e ™- 



tvi 



(when the anode load resistances i? : and R 2 are of equal 



value and the internal resistances of the valves are equal) and the 
17 



226 



TIME BASES 



unbalance of the resistances R 3 and i? 4 = BR = 



*4 



*S 



i? S 



with 



R e = |(i? 3 + /? 4 ). The curves are expressed as percentages for different 

values of the gain ,4 = — 2 of the amplifier stage V 2 , the internal 
R a2 + K 2 

impedance of which is denoted by R a2 and the amplification factor by u.. 

Thus, if the gain A is 40, BR must be +7-5 per cent for zero output 

unbalance. For perfect balance with equal anode loads and the values 

of R a equal, BR must equal $'<A approximately. 

The values of the resistances forming the anode loads R x and R 2 
are chosen in the ordinary way and, since high gain is required, valves 
having a reasonably high value of internal impedance should be 
employed. The choice of values for R lt R 2 and R a are, of course, 
subject to frequency response requirements. Since the two valves have 
to perform similar functions it is advisable to employ two of the same 
type or, alternatively, a double triode may be used. 

An advantage of this circuit, and of that next to be described, is that 
ripple is not introduced on to the H.T. supply rail by the circuit. 

An Amplifier combining the Principles of Cathode and Anode 
Coupling 

The principles of the anode and cathode coupled circuits may be 
combined, with the result that the degree of balance, once it has been 
adjusted, is held more closely. 

The circuit is shown in Fig. 156, from which it will be seen that the 
cathode load resistance R s is unshunted, so that the grid of V 2 is driven 
partly by variation of the potential of its cathode and partly by the 
application of a portion of the potential change at the anode of V x to 
its grid via C lf R z and R 6 . These two potentials are, of course, in 
phase opposition but, since they are applied to the cathode and the 
grid of V 2 , they aid one another in producing a potential change at the 
anode of V 2 . 

By placing the balancing resistances R s and R A after the necessary 
time constant circuits C 2 R 6 and C%R 7 which feed the following push- 
pull stage (as shown in Fig. 157), any small inequality of these two cir- 
cuits can be compensated provided that their impedance is high com- 
pared with that of the anode load resistances R 1 and R 2 at all frequencies 
within the required band. 

As with the see-saw circuit, the gain of the F 2 stage should be made 
as high as possible and, of course, pentodes may be employed instead 
of triodes. The value of the resistances 7? 3 and i? 4 is immaterial except 
that they should not be so low in value as materially to reduce the gain 
of the two valves since, of course, they appear, individually, in parallel 



PUSH-PULL DEFLECTION 



227 




R3 



Output 



X.4 



MMW WW WW- 

HI 



Ci 



Input 



\Rs 




OH.T.+ 



-0//.7T- 



Fig. 136. An amplifier in which the principles of anode and cathode coupling are 

combined 




Rb 



Output 



1L 



|--WW VW WW-— I 



C3 



Input 



*7 



: *5 




-o//.7> 






-0//.7:- 

FlG. 157. An amplifier which is similar to that of Fig. 156 but in which the low- 
frequency errors in the coupling networks C 2 R g and C 3 7? 7 are compensated 

A~cc 

100 
60 
40 

2D 



10 




-70% 



^ -20% 



-30% 



-40% 









































A 


y 











































~30% -20% -10% +10% +20% +30% 

Fig. 158. Relation between the output unbalance Be and the resistance un- 
balance BR for different values of gain A of the V ., amplifier stage of Fig. 157 
when Ri and R & are of equal value and Rat equals Ra% 



228 



TIME BASES 



with the anode load resistance of the valve to which they are each 
connected. Fig. 158 shows the percentage error in balance of the output 
potentials for the circuit of Fig. 157, with equal values of anode resistance 
i?j and R 2 and equal values of R a , for various percentage differences 
in the values of the resistances R 3 and i? 4 . For perfect balance with 
equal anode loads and R a equal for the two valves, BR must be approxi- 
mately equal to 2/ A. These curves differ from those of Fig. 155 because 
the shunting effect of the grid leak resistance is absent. 

The fixed parts of the resistances i? 3 and # 4 may be replaced by 
condensers.* 

PROVISION OF PUSH-PULL DEFLECTION WITHOUT THE 
USE OF AMPLIFIERS 

(d) Split Time Bases 
Since a saw-tooth wave-form is difficult to amplify without the 
introduction of distortion, it is often desirable to avoid the necessity 
for amplification by generating a saw-tooth wave-form having sufficient 

-o/iJ.-f- 

Cz 



o//.r- 



Fig. 159. Split time base employing a single capacitance 
and two charging devices 

amplitude to allow half the output potential to be applied to each 
deflector plate. The methods illustrated in Figs. 159 and 160 indicate 
the principle involved. Either the time base condenser or the charging 
resistance may be composed of two equal sections so that balanced 
output potentials of opposite phase become available for application 
to the deflector plates of a cathode ray tube. Fig. 159 shows a circuit 
in which two charging devices of equal impedance (here shown as the 
resistances R 1 and R 2 ) have a condenser C t between them. The 
discharging device is indicated by a key. It will be obvious that one 
plate of the condenser will become more positive while the other 
becomes more negative. The condensers C 2 and C 3 and the resistances 
R z and R 4 are inserted to allow the sweep to centre itself on the screen 
by isolating the deflector plates from the H.T. supply potential of 
the time base, in order that one end of the supply may be connected 
to an appropriate point (say, to earth) in the H.T. potential system 

*C. L. Hirshman: British Patent 491468. 







PUSH-PULL DEFLECTION 



229 



of the cathode ray tube. The lower ends of the resistances R 3 and R 4 
are therefore shown connected to earth, to which point the final anode 
of the cathode ray tube is usually connected. 

Dodds' Time Base 

Fig. 160 shows a circuit* in which only one charging device is 
employed, but the condenser is divided into two equal portions C^and C 2 . 
By connecting the junction of the condensers to earth, the potentials 
across them vary in opposite phase with respect to earth. The resis- 
tances 7? L and R 2 > which may be of the order of 5 megohms each, are 
inserted to enable the mean potential of the pair of deflector plates to 
be fixed with respect to the final anode of the cathode ray tube which 



Xi 



Ci 



T Q 



O/i.T.-h 



7 K 



X-Shift 



vifc 4 ^ 



Fig. 160. Dodds' split time base employing two 
centre capacitances and one charging device 

is connected to the junction of these resistances. The pentode V lt 
whose anode current may be controlled by adjustment of the slider 
of R i} is employed as a constant-current device but, if greater linearity 
is required, a cathode feed-back resistance may be inserted, provided 
the potential of the grid is suitably adjusted as explained in Chapter VIII. 
A discharging device, represented by a key, serves to reverse the charge 
of the condensers C 1 and C 2 . The effect of adjusting the slider of R s is 
to alter the initial position of the spot on the cathode ray tube screen 
and, hence, to produce a horizontal shift of the trace. The condenser 
C 3 is made large enough to hold the time base H.T. supply at a constant 
potential difference from the X 2 plate. 

The fact that the upper plate of the condenser C x is connected, 
during the charging regime, to a potential which is lower than the 
maximum obtainable, does not reduce the resultant deflection. This 
is because, during discharge, the lower plate of C s , instead of being 
brought to the same potential as the upper plate of C 1} is returned to 
a potential which is positive with respect to it. In this way the trace 
is made to start from a rigidly fixed point near the edge of the screen, 
with the advantage that the trace does not "float" if the repetition 
frequency is not constant. A minor disadvantage of this circuit is that 

* E. M. Dodds: "Recent Developments in Engine Indicators," Journ. Inst 
Auto. Engineers, 1937. 6, p. 41. 



230 



TIME BASES 



the time base H.T. supply must be isolated from earth, any leakage or 
capacitance to earth tending to upset the balance. 

(e) Two Equivalent Time Bases Operating in 
Phase Opposition 
Fig. 161 shows a circuit which, in essence, consists of two time bases 
of which one is connected in the reverse sense to the other. In this 
circuit, a single discharging device may be employed but, with this 
arrangement, the H.T. potential requires to be greater than is necessary 
if separate discharging devices are employed. R l and R 2> here shown 
as resistances, may be equal constant-current devices, and C l and C 2 
are equal condensers which are charged in opposite sense. The key 
which represents the discharging device is used to bring the condensers 
to the same potential and, since R x and R 2 are of equal impedance, 

-ott.J.-t 




-o/zr- 

Fic. 161, Basic circuit of two time bases operating in push-pull 

this potential is half that of the H.T. source. When the key is opened, 
the X 2 deflector plate is driven in a positive direction and the X x plate 
negative. The deflector plates are isolated by the condensers C 4 and C 6 . 

Fig. 1 62 shows how two time bases may be employed to provide a push- 
pull output so that almost the whole of the available supply potential 
may be made to appear across each of the condensers Cj and C 2 . 

In this circuit, the two gas-filled discharge valves V x and V 2 are 
connected across the condensers C 2 and C 2 respectively, and the bias 
potentials applied to their grids are determined by adjustment of the 
resistances R b and i? 6 . The potentials across these resistances remain 
steady because the ripple is removed by the condensers C 4 and C 5 , 
whose capacitances are such that 

C A R,>T 
and C 5 R 6 >T, 

where T is the longest required duration of the time base trace. 

It will be noticed that, while the condensers C x and C 2 are being 
charged, the potential of the anode of V 2 remains fixed, while those 
of the cathode and grid rapidly become more negative with respect 
to the anode. Since the grids of the two valves are joined together via 
the condenser C 3 , and the discharge current of C 3 passes through R 2 , 



PUSH-PULL DEFLECTION 



231 



the grid-cathode bias of V 1 is greater than that due to the potential 
drop across R s . For this reason, R 6 must be adjusted to provide a 
considerably smaller bias potential than is the case with i? 6 since, 
otherwise, there will be a time lag between the ignition of V 2 and that 
of V x . It will be obvious that unless *y? 6 is equal to or greater than 
i 1 R 5 plus the added potential drop across R 3 (due to the fall of potential 
at the grid of V 2 during the trace period), F 2 must fire first. V lt how- 
ever, cannot fail to fire eventually because, as soon as V 2 fires, the 
potential of the cathode and grid of V 2 rapidly becomes more positive 
as C 2 discharges and the grid of V x is also driven positive by the 

I -f 1 o/i.r-t- 

1 i*s 




OSyrtch. 



0//T' 



Fig. 162. Two thyratron time bases operating in push-pull 

charging current of C 3 which passes through R z . The object is to 
obtain simultaneous ignition, however, and this is best achieved by 
making the potentials at the two grids equal at the instant immediately 
prior to ignition by accurate adjustment of R 5 . 

Suggested values for the components are as follows : 
R t and R 2 0-5 to 5 megohms each (equal in value). 
■^3 » ^4 50,000 ohms each. 
R b „ R 8 a little more than 4 per cent of R x each, for valves 

having a control ratio of 25. 
R 7 „ R 8 500 to 1,000 ohms each. 
-^9 » ^io 2to 5 megohms each (equal in value). 
C 8 generally not more than 50 to 100 iifxF. 
The values of the condensers C x and C 2 naturally depend upon the 
required rate of traverse of the cathode ray spot and, obviously, R 1 and 
R 2 may be replaced by constant current- devices if desired. 
The values of C 7 and C 8 should be such that 

C 7 i? fl > T 
a "d C S R 10 >T. 



THE SYNCHRONIZATION OF TIME BASES 



233 



CHAPTER XI 
THE SYNCHRONIZATION OF TIME BASES 

THE GENERAL PROBLEM 
The problem of the synchronization of time bases is one of the greatest 
importance. Synchronization is the timing of two events to coincide 
with one another. When a time base potential or current is to be 
synchronized with an external event, the effect is produced by the 
application of a signal in such a way that it initiates either the sweep or 
the fly-back of the time base. It is more usual, and certainly easier, to 
initiate the sweep with the synchronizing signal than to initiate the 
fly-back. However, in most precision apparatus, and in many radar 
applications, it is essential to initiate the fly-back. Thus, in the case of a 
capacitive time base, the synchronizing signal initiates the current in 
the discharge valve or, alternatively, it may cause this current to be cut 
off. In the case of a blocking oscillator, it initiates the flow of anode 
current which is then maintained by oscillator action. When a syn- 
chronizing signal is applied to a multivibrator or a flip-flop, it initiates 
a change of regime by starting or arresting the flow of anode current 
in one of the valves. 

A synchronizing signal should, in general, have a steep wave-front 
in order to give precision of time control and, for the same reason, it is 
advisable, where possible, to apply the signal to such a part of the time 
base, or other circuit, that it becomes amplified before reaching the 
point at which it exerts the synchronizing effect. Synchronization can 
be effected with any wave-form, subject to the polarity of the signal 
being suitable, but a steep-fronted pulse is the most effective form of 
signal. 

If the time base has a trigger device which operates automatically 
at the end of each sweep, the time base is essentially self-running and 
the synchronizing signal is preferably, but not necessarily, completed 
before the end of the fly-back period. If the time base circuit can 
ignore the presence of the signal after the completion of the fly-back, 
there is obviously no necessity for restricting the duration of the 
synchronizing signal, and many circuits operate in this condition. In 
the absence of a synchronizing signal, the time base continues to func- 
tion, but the duration of the trace is not necessarily the same. Sometimes 
the synchronizing signal, or pulse, is fed via a biased diode to an 
electrode of the trigger valve and the resulting trigger action is arranged 
to remove the bias from the diode so that no further signals can affect 
the circuit until it has returned, once more, to a quiescent condition. 

232 



I 



When the time base contains no automatically operating trigger device, 
the signal should not be considered as a synchronizing signal in the 
strict sense of the term since, if the time base cannot operate without 
the signal, the latter cannot truly be said to synchronize but rather to 
control the operation. In this case also, the wave-front should pre- 
ferably be steep to provide precision of timing, but the duration of 
the control signal (generally in the form of a pulse) in some cases deter- 
mines the interval between the commencement of the trace and of the 
fly-back, or vice versa, according to the polarity of the signal. Some- 
times a time base, or other circuit, has two controlling signals of w*hich 
one serves to initiate the sweep and the other to initiate the fly-back. 
However, many externally operated time bases contain trigger circuits. 
If there is no trigger circuit, the commencement of the trace requires 
an initiating signal and the commencement of the fly-back requires 
another, A time base which has no trigger device is thus an externally 
operated one and, for its operation, the existence of a control signal is 
essential. When the time base includes a trigger device of the flip-flop 
type, the same remarks apply except that the duration of the sweep is 
normally independent of the duration of the control signal. 

In order to achieve accurate synchronization or control of a time 
base, it is essential that only the required signal be permitted to exercise 
any measure of control upon the circuit. This implies that: 

(a) the high tension supply to the time base shall be steady in value 
and quite free from ripple, 

(h) no hum modulation shall be introduced from the cathode heating 
circuits, 

(c) the synchronizing electrode shall not pick up any extraneous 
varying potentials by stray capacitance, or in any other way, 

(d) no magnetic modulation of the currents in the valves or induc- 
tances shall take place, 

(e) the wave-front of the synchronizing signal shall be steep enough 
to swamp the effects of any random potentials which may occur 
at a vulnerable point in the circuit, 

(/) the synchronizing signal must be free from any irregularity; in 
particular, no interference occurring at frequencies which are not 
harmonics of the fundamental synchronizing frequency can be 
tolerated since, in this case, the interference will glide through 
the signal and cause errors in timing, 

(g) the time base must be relatively free of inherent frequency drift, 

(h) the synchronizing signal or pulse must be of sufficient amplitude 
and, in some circuits, it must not be of too great amplitude. 

It is often inconvenient to provide only one synchronizing pulse 
during each trace. If there are many pulses during one trace, the 



234 



TIME BASES 



inherent short-term frequency stability of the time base must be good 
or the number of pulses per trace may vary. This will, in turn, be likely 
to cause fluctuation of the delay between the commencement of the 
synchronizing pulse and the trigger action. In such cases, the ampli- 
tude of the pulse will usually be found to have an optimum value which 
will provide the most satisfactory result. In those cases where there 
are many synchronizing pulses per trace, a biased diode circuit which 
has its bias potential removed by the action of the trigger valve, as 
described on page 250, may be employed (see also page 256). 

If the synchronizing pulses are less frequent than one in each trace, 
as sometimes occurs with a television receiver when fading conditions 
are bad, pulses may be applied to control the average repetition fre- 
quency of the time base in such a way as to maintain a constant average 
time relationship between the time base and such pulses as are received. 
In this system, the individual traces are not synchronized, but the 
natural frequency of the time base is automatically corrected upon the 
arrival of a synchronizing signal with which the time base is slightly 
out of step. This requires that the time base must be adjusted very 
much more closely to the correct operating frequency than is normally 
necessary. 

When the wave-form under observation varies in time from cycle 
to cycle, synchronization in the ordinary manner is not very helpful 
This problem is particularly acute in the application of the oscillograph 
to measurements on internal combustion engines. Owing to slight 
variations in the ignition point and in the quantity of fuel drawn into 
the cylinder, the time taken for each complete cycle varies slightly but 
sufficiently to produce a blurred diagram on an oscillograph which 
has a constant-velocity time base. Only a small part of the diagram 
can be held quite steady, and this is the part immediately following 
the synchronized fly-back. The complete diagram can be examined in 
detail, step by step, if the synchronizing pulse can be made to occur at 
successive points in the rotation cycle of the engine. For this reason, 
the time base is preferably synchronized by means of a pick-up coil 
mounted on the engine in such a way that a small projection on the fly- 
wheel will induce a pulse at a precise point in each revolution. Means 
are usually provided for moving and clamping the pick-up device at 
any desired angular position. The same result can be obtained by 
using a cam-operated contact maker, which is adjustable so that it may 
be operated at any point in the cycle of revolution, to discharge the time 
base condenser.* 

* In the case of four-stroke engines, it is an advantage also to mount on the timing- 
shaft a commutator arranged to interrupt the circuit to the contact maker on alternate 
revolutions, so that each time base trace corresponds to one complete engine cycle of 
two revolutions. 



THE SYNCHRONIZATION OF TIME BASES 235 




An alternative solution is r oh.t.+ 

to derive the horizontal sweep 
potential from a device which 
is coupled to the engine shaft 
and arranged to produce a 
potential which is directly 
proportional to (a) the angle 
of revolution, or to (b) the 
piston displacement. A wire- 
wound potentiometer having 
a rotating contact arm may be 
used for this purpose; the 
resistance law should be linear 
for (a) or suitably graded for 
(b). In a case of this sort it is probably advisable to arrange for the 
engine to operate a switch to provide a short circuit which is removed 
for a short time at an adjustable point in the cycle of operation. The 
switch may operate a simple charging circuit of the type shown in 
Fig. 163. This type of circuit is very stable in operation, particularly 
when the condenser is held discharged during the waiting period. 

, Modified Striking Potential 
when sinusoid is applied to 
the grid of the thyratron 



Fig. 163. Simple circuit for making measure- 
ments on internal combustion engines. The 
switch, which is operated by the engine, is 
normally short-circuited 




Time 



Fig. 164. Effect of applying a sinusoidal synchronizing poten- 
tial to the grid of a thyratron (synchronization ineffective because 
the tube does not fire at similar points on the synchronizing 
wave-form) 

Synchronization of a Soft Valve Time Base 

The synchronizing signal is applied to the grid of the thyratron or gas- 
filled discharge tube, and has the effect of altering the value of the anode 
potential at which the vapour or gas ionizes. Fig. 164 shows the effect 



236 



TIME BASES 



produced when a sinusoidal synchronizing potential is applied to the 
grid. In the diagram, A shows the ionizing or firing point in the 
absence of the signal, and B shows the modified striking potential when 
a sinusoid is applied to the grid. It will be obvious that it is sometimes 
impossible to synchronize the time base when it is operating too fast. 
This does not mean that the time base can never be synchronized if 
the natural repetition period is less than that of the synchronizing signal. 



Modified Striking Potential 
with sinusoid applied to 
grid of thyratro/} 

A 



Striking Potential 



FlG. 165. Effect of applying a sinusoidal synchronizing potential to the grid 

of a thyratron (synchronization is effective because the natural period is longer 

than a multiple of the synchronizing period) 



Fig. 165 shows how a time base can be synchronized when the 
natural period is slightly longer than the synchronizing period or a 
multiple thereof. It should be noted that the natural period is not 
complete until the point A is reached. This is the point at which the 
trace ceases in the absence of a synchronizing signal. 

From the foregoing remarks it will be seen that when the trace 
potential line reaches the synchronizing potential at a point above that 
at which the tube would strike in the absence of a synchronizing signal, 
the time base period is increased, and when it strikes below this potential 
the period is reduced. 

Examination of Fig. 165 makes it obvious that the larger the number 
of synchronizing signals occurring during the period of one trace, the 
smaller must be their amplitude since otherwise the trace potential may 
cross the bottom of the previous cycle and, in this case, the trace length 
will be reduced. 




THE SYNCHRONIZATION OF TIME BASES 



237 



Builder and Roberts* have published useful curves showing the 
permissible ratios of the natural repetition frequency to the control 
frequency. These curves are reproduced in Figs. 166, 167 and 168. In 
these curves, fojf is the ratio of the natural free-running repetition 
frequency of the time base to that of the sinusoidal control frequency, n 




7-4 



i-Z 



1-0 



0-8 



\ 



0-6 



04 



0-2 





X 


















( 

\ 
\ 
s 

\ 






x\ 




i\ 
\ 
\ 

\ 
\ 


/7=; 




\ 




oS. 






*\ 


Ox 


> 

s 
N. 


o\ 


*\ 






V 


o^S.\ 



01 0-4 0-6 
Fig. 166. 



;0 



0-2 0-4 



0-6 

SJ£ 
Fig. 167. 



0-8 1-0 



FlG. t66. Permissible ratios of the natural repetition frequency to the control fre- 
quency, shown by shaded areas, for synchronization at the Hth submultiple of the 
control frequency (exponential charge! 

FlG. 167, Maximum (full line) and minimum (broken line) ratios of the natural 
repetition frequency to the control frequency for synchronization at submultiples 
n = i and n^2 of the control frequency corrected for the varying rate of charge of 
the time base condenser. Experimental results are shown hy crosses («-~i) and 

by circles (n 2) 

is the ratio of the synchronized time base frequency to that of the con- 
trol frequency and EJE is the ratio of the synchronizing potential 
appearing at the anode of the thyratron (i.e. the peak grid potential 
multiplied by the control ratio) to that of the potential at which the 
tube would strike in the absence of a synchronizing signal. 

* G. Builder and N. F. Roberts: "The Synchronization of a Simple Relaxation 
Oscillator," A.W.A. Technical Review, 1939, 4, p. 165; O. I. Butler: "Synchronization 
of Controlled Relaxation Oscillators," Phil. Mag., 1948, 39, p. 518. 



2 3 8 



TIME BASES 



THE SYNCHRONIZATION OF TIME BASES 



239 



Fig. 167 relates to a linearly charged condenser, the H.T. potential 
being 250 volts and E being 90 volts. 



$ 



rz 

f-0 










































0$ 

0-6 


* 


■s 


























4> 








0-4 






















0-2 







































^^^ 





Of 



0-2 



0-3 



0-4 



0-6 0-7 



0-8 



09 



r-o 



OS 
E,/E 

Fig. 168, Permissible ratios of the natural repetition frequency to the nth sub- 
multiple of the control frequency for synchronization at that subrnultiple. In 
each case the minimum permissible ratio is shown by the broken line curve. The 
maximum permissible ratios are shown by the full line curves for several values of n 

In these curves, several interesting features appear : 

(a) When w/ =/, synchronization can only occur for values of EJEq 
less than a quantity approximately equal to 0-73/7*, for example 
0-73 for ?j=i and 0-15 for 72=5. 

(b) For values of EJEq greater than about 073/M, the natural fre- 
quency /„ must be less than the locked frequency //«, if locking 
is to occur. 

(c) The locking regions do not overlap for any values of EJE . For 
this reason, successive conditions of synchronization are separated 
by regions where synchronization ceases to occur as EJE Q or 
fjf is varied. 

Further details concerning the synchronization of a soft valve time 
base appear on page 19 et seq. 

Synchronization of a Soft Valve Frame Time Base in a Television 
Receiver 

Tomlinson, Hosking and Schnetzler* have shown that, in order to 
prevent "pairing" of the scanning lines in an interlaced raster when 

*T. B. Tomlinson, J. Hosking and K. Schnetzler: "A Problem of Interlacing with 
Thyratron Time Bases," Jottrn. Television Society, 1947, 5, p. 595. 






employing a soft valve time base, having differentiated frame syn- 
chronizing pulses, the following precautions should be taken : 

(a) The frame fly-back should be completed within a period of 400 
microseconds in order to prevent any effects due to residual line 
pulses in the frame synchronizing signals. 

(b) The arrangement of the frame scanning oscillator should be such 
as to obtain the minimum potential pick-up from the line- 
scanning output valve, at the grid of the frame-scanning thyratron. 

(c) If pick-up is, for any reason, unavoidable, the frame fly-back 
time and, hence, the value of the discharge current limiting 
resistance (see page 22) in the anode circuit of the thyratron, 
should be so adjusted that the commencement of a new frame 
scan does not occur within some 20 microseconds from the 
leading edge of a line synchronizing pulse. This, of course, applies 
to each frame scan of the completed picture. 

Keen* has discussed the frame time base synchronization of tele- 
vision receivers in considerable detail and his paper is of very great 
value. 

Synchronization of a Hard Valve Time Base 

In the case of a hard valve time base, the synchronizing signal brings 
about a change of regime by cutting off, or on, the anode current of a 
valve by the application of a suitable signal to another electrode. In 
the case of the author's time base, shown in Fig. 23, the anode current 
of V z is reduced to zero by the trigger action which results upon the 
arrival of a synchronizing signal. The cyclogram of Fig. 32 is repro- 
duced here, together with a new one, to show the difference which 
appears when the synchronizing signal is applied. In the cyclograms, 
Fig. 169 (a) shows the result obtained in the absence of a synchronizing 
signal, while (b) shows five synchronizing signals (in this case, sinusoids) 
per trace. It will be seen that the signal applied to the grid of the 
pentode F 3 appears also on the grid of the discharge valve, but in 
amplified form and reversed in polarity, where it serves to initiate the 
discharge of the condenser by advancing the time of commencement of 
anode current in V 2 . As the synchronizing signal is increased in ampli- 
tude, the synchronizing effect takes place earlier, jumping suddenly 
from each cycle to the preceding one. This effect is very valuable and 
indicates the existence of extremely "tight" synchronization. 

Note that the time base trace was not linear when Fig. 169 was photo- 
graphed. This is made evident by the varying time scale of the 
sinusoidal wave-form at the foot of Fig. 169 (/>). 

* A. W. Keen: "Frame Time Base Synchronization in Television Receivers," Jourtt. 
Television Society, 1947, 5, pp. 49 and 124. 



240 



TIME BASES 



A method of keeping a blocking-oscillator operating in such a way 
that the amplitude of the saw-tooth wave-form is automatically re- 
adjusted to a predetermined value even if the repetition frequency of 
the synchronizing signals should vary has been invented by Faudell 
and White.* 




*S V z 

Anode ■ Grid 
Volts Volts 




a) Without synchronizing signal 



(b) With synchronizing signal 



Fig. 169. Effect of the application of a sinusoidal synchronizing potential to 
control electrode of V % in Fig. 23 

Synchronizing -pulse Generators 

Synchronizing-pulse generators form an essential part of a television 
transmitter and they are also used for certain other purposes. When 
used as part of a television transmitter \ they must be arranged to take 
full control of the output of the picture transmitter during those periods 
when a synchronizing pulse is required to be transmitted. For this 
result to be achieved the synchronizing signal must perform two 
functions : 

(a) It must prevent the passage of any picture information to the 
transmitter, by cutting off the modulation due to the picture for 
the period of the synchronizing pulse. 

(b) It must modulate the transmitter to substantially zero or maxi- 
mum output (according, respectively, to the British or American 
standard) for the period of the synchronizing pulse. 

If this suppression of the picture modulation is not perfectly achieved 
accurate synchronization of the received picture becomes very difficult 

*C. L. Faudell and E. L. C. White: British Patent 491934. 

t A. V. Bedford and J. P. Smith: "A Precision Television Synchronizing Signal 
Generator, R.C.A. Review, 1940, 5, p. 51; R. A. Monfort and F. J. Somers: 
Measurement of the Slope and Duration of Television Synchronizing Impulses " 
R.C.A. Reviezu, 1942, 6, p. 370. ' 



THE SYNCHRONIZATION OF TIME BASES 241 

because the edge of the synchronizing pulse will fall (or rise) at varying 
rates due to the super-imposition of picture modulation. In practice, 
it is found necessary to bring the level of the carrier to "black" about 
£ microsecond before the insertion of the synchronizing signal, in order 
to provide a common starting level for all synchronizing signals, and so 
to keep the rate of fall of carrier amplitude constant. 

This short period, when the carrier is at black level, is known as a 
pedestal. It is common practice to provide a further short period at 
black level after the completion of the synchronizing signal and before 
the picture modulation is again transmitted. 









18 



CHAPTER XII 

THE USE OF A TIME BASE FOR FREQUENCY DIVISION 
AND FOR COUNTING 

A time base may be used for time measuring purposes without the 
application of the generated potential to the deflector plates of a cathode 
ray tube. For example, a time base, having a loud speaker connected 
to its output terminals, has been used by the author to serve as a 
metronome. Perhaps the most important application of this kind is to 
the division of frequency and the counting of impulses. Frequency 
division and counting are distinct operations and should not be con- 
fused. Table VI gives a family tree of time bases used without a 
cathode ray tube. 

Table VI 

Time Bases 



Frequency 
dividers 



Stable 
triggered 
integrators 



Triggered 
integrators 
having only 
short-term 

stability 



Synchronized Synchronized 
stabilized time bases 
time bases 



Miscellaneous 
applications 



Stable 
time bases 



Self-running 
or unstable 
time bases 



Counters 

I 

Stable 

triggered 

integrators 



Time bases for use without a cathode ray tube 

FREQUENCY DIVISION 

Frequency division involves the apparent integration of a number* 

of input pulses or cycles whose quality and temporal spacing is definitely 

specified within close limits. Thus, the input to a frequency divider 

may be of sinusoidal, saw-tooth or pulse form, but, on the other hand, 

the input wave-form must remain constant within very narrow limits. 

Frequency dividers may be stable or semi-stable externally triggered 

time bases, but they must remain stable for a period greater than that 

•This number may be any integer including i and generally, but not always 
limited to a maximum of 10. 



242 









. 



USE OF TIME BASE FOR FREQUENCY DIVISION 243 

of the division period, i.e. one output period. Another form of frequency 
divider (Builder's Circuit) includes a time base from whose output a sine 
wave is generated by a tuned circuit and this sine wave is used to control 
the time base repetition frequency. Any time base or trigger circuit 
which is operating in a condition of stable synchronization, usually at a 
repetition rate which is a submultiple of the input frequency, is acting 
as a form of frequency divider. 

In essence, a frequency divider will only function correctly when the 
input frequency and wave-form can change only over narrow limits, 
but certain dividers will continue to function correctly when the 
input frequency is changed over fairly wide limits. Furthermore, the 
circuit must return to a stable condition before being re-triggered since, 
otherwise, a variable phase condition will arise between the input and 
output wave-forms. 

Some frequency dividers integrate the input pulses or cycles of the 
input wave-form and, when a certain potential appears across the 
integrating (time base) condenser, the discharge circuit is triggered and 
the circuit is restored (see page 254). As a result the wave-form 
appearing across the integrating condenser is scalariform.* This wave- 
form also results from the racheting process described on p. in, and 
in the counting process described on p. 257. 

A frequency divider must not be liable to inherent frequency drift. 
Even if the circuit is not liable to frequency drift, the ordinary time 
base is, in general, not ideal for use as a frequency divider because the 
frequency ratio (input to output) depends on several adjustments which 
are difficult to calibrate when a high dividing ratio is required. More- 
over, these adjustments are sometimes interdependent. For small 
dividing ratios and long time periods, i.e. low frequencies, it is relatively 
easy to adjust a normal time base to function as a frequency divider. 

Frequency division may be carried out in stages so that large division 
ratios may be obtained while allowing each stage to operate with a 
division ratio which is not so large as to result in imperfect synchroniza- 
tion. Frequency dividers are sometimes called counting down circuits. 

Frequency Division from a Sinusoid 

It is possible to "divide" a sinusoidal source in order to obtain 
another source (output) at a lower frequency which is harmonically 
related to the "master" frequency of the original source. This may be 
achieved by means of a time base locked to the original source with a 
means of converting its output wave-form to the sinusoidal shape. 
Buildcrf has developed circuits for achieving this result. 



• Scalariform — like the succession of steps of a staircase. 

t G. Builder: "A Stabilized Frequency Divider," Proc. I.R.E., 1041, 



h p. 177. 



244 



TIME BASES 



Builder's Frequency Divider 

If the synchronized time base is not inherently stable, there is a 
possibility that it will not always "lock" to the desired sub-harmonic 
of the synchronizing signal. For example, if the master oscillator fre- 
quency is 100,000 cycles per second and the desired low frequency is 
1,000 cycles per second, the system may wander from the 100th sub- 
harmonic to the 99th, which is 1,010-1 cycles or to the 101st, which is 
990-1 cycles per second. In order to guarantee that the output frequency 
shall be 1,000 cycles per second, it is essential, in this case, to make the 
low-frequency oscillator incapable of oscillation outside a range of, say, 
991-1,009 cycles, i.e. stabilization of the frequency must be obtained/ 
In attempting to synchronize a time base from a sinusoidal source 
of much higher frequency in order to provide an oscillograph sweep, 
a similar state of affairs is encountered, but the frequency ratio may 
be varied considerably without disturbing the appearance of the image, 
except that the horizontal scale is slightly changed. Stability of the 
time base frequency over long periods of time is therefore quite un- 
important m this case, although the stability over short periods should be 
reasonably good, otherwise the frequency ratio will only remain constant 
for a short tune and will then jump to another value. A similar effect 
takes place if the amplitude of the synchronizing signal is not constant. 
The use of the words "synchronization" and "stabilization" is 
intended to convey that synchronization provides rigid locking of the 
system to any sub-harmonic frequency whereas stabilization provides 
limitation of the frequency range over which the system can be locked. 
Thus, stabilization plus synchronization provides rigid locking to a 
specific sub-harmonic frequency. 

Koga,* Andrewf and Builder have all discussed the use of re- 
generative oscillators which are inherently capable of oscillating only 
at frequencies near that of the required sub-harmonic, and Builder 
has succeeded in developing a circuit of this type in which the degree 
of stabilization is sufficient to enable the high-frequency driving source 
to hold the circuit oscillating at a frequency which is rigidly under the 
control of the source. 

The problem of the synchronization of an oscillator or generator is 
greatly eased if its wave-form contains a large number of harmonics. 
Lor this reason, therefore, a multivibrator or saw-tooth time base is 
a suitable form of generator. The reason for the greater success of 
Builder's circuit, by comparison with those of Koga and Andrew, is 

MM* %£*''££ NCW FrCquenCy Tra "sformer or Frequency Changer," Proc. I.R.E., 

t V.J. Andrew: "The Adjustment of the Multivibrator for Frequency Division M 

lTB.;%fi 21? P- 9 8' 2 P ' 19 " : " A Simplified FfeqUenCy Dividin * CirLk> £"' c . 



USE OF TIME BASE FOR FREQUENCY DIVISION 245 

that the former employs a stabilized relaxation oscillator which can 
only function over a small frequency range, whereas the others use 
multivibrators or ordinary oscillators which are modulated in a normal 
manner and no special precautions have been taken to limit the range 
of frequency over which the circuits can operate. Builder's circuit is 
of especial interest in the present circumstances because it employs 
a time base as an essential part of the circuit. 

OHT+ 

Out 



Control 
Potent/a/ 
o 




oH.T- 



FiG. 170. Builder's circuit for frequency division by means of a time base 

The principle upon which Builder's stabilized frequency divider 
provides the required "tight" synchronization at the correct frequency 
is that of the association of a relaxation oscillator with feed-back through 
a frequency-selective network. This network, which is tuned to the 
frequency of subdivision, provides stabilization of the output frequency 
within very close limits and the output of the tuned network is fed to 
the grid circuit of the relaxation oscillator via a phase-changing circuit 
(see Appendix VII). The master oscillator then synchronizes the whole 
system to the precise frequency required. In this way, Builder's circuit, 
which is shown in Fig. 170, is constrained, by the tuned circuit L t C 4) 
to oscillate only at a frequency within a narrow range. Since this 
circuit is tuned to the required sub-harmonic of the controlling fre- 
quency, it is impossible for synchronization to occur at any other sub- 
harrnonic frequency. 

In the circuit of Fig. 170 the relaxation oscillator, or time base, circuit 
includes the valve V x which discharges the condenser C v the latter 
being arranged to acquire a charge via the resistance R z . The resistance 
R± is included for the purpose of preventing the flow of excessive 
discharge currents in the gas-filled triode V 1 (see page 22). Obviously 
V 1 may be replaced by a hard valve trigger discharge circuit if desired. 
The anode circuit of the valve V 2 is tuned to provide the selection of 
the appropriate frequency. One of the secondary windings of the 



246 



TIME BASES 








(b) 




transformer T z supplies the feed-back potential, the phase of which 
is adjustable by altering the value of the condenser C 6 . The other 
winding provides the sinusoidal output potential. 

The time constant C^ and the bias resistance R 2 are preferably 
adjusted, in the absence of feed-back, so as to cause the natural repeti- 
tion frequency to be slightly less than the required output frequency. 

The resistances R x and R 5 are 
each of high value in order to 
prevent the grids V x and V 2 from 
becoming substantially positive 
with respect to their cathodes. 
The resistance R x must not be 
too high since, othenvise, the 
application of the negative poten- 
tial to the grid will not extinguish 
the discharge. With a hard-valve 
trigger circuit this remark does 
not apply. 

Fig. 171 shows the wave-form 
existing at various points in the 
circuit in the absence of the con- 
trol potential which is normally 
applied to the primary winding 
of the transformer T v The time 
base wave-form is shown in the 
diagram as an ideal linear saw- 
tooth for the sake of convenience. 
The small vertical part of the 
otherwise sinusoidal wave-form is 
a reflection of the fly-back of the 
time base which appears in the output by stray capacitance coupling 
through the valve and transformer. Starting with the output wave- 
form shown at (a); this is applied via the phase-shifting network to 
the grid of V x where it appears as in (6), the upper portion being 
removed by the action of grid current flowing through the resistance 
J?* The commencement of the charging cycle (c) of the condenser C x 
is delayed until the grid potential of the valve V x has been reduced by 
the sinusoid to a value which is sufficiently negative to cut off the 
anode current and so to allow the condenser C x to charge once more 
It should be noted that the condenser C x is residing in a discharged 
condition, i.e. the potential across the valve V x is the ionization 
potential for the particular vapour, or gas, with which the valve is filled. 
In this condition, the application of a negative potential to the grid is 




id) 

Fig. 171. Wave-forms relating to 
Builder's frequency dividing circuit 






USE OF TIME BASE FOR FREQUENCY DIVISION 247 

able to cut off the flow of anode current provided that the anode 
current is very small. In order to ensure precision of operation, the 
amplitude of the sine wave applied to the grid of V x should be at least 
five times the grid base of the valve. The grid base referred to here 
is the potential required to cut off anode current (i.e. to extinguish 
the discharge) with the condenser in the stable discharged condition. 
In these circumstances, the presence of the condenser prevents the 
anode potential from rising rapidly while the grid potential is being 
reduced. The grid base is likely, in this condition, to be of the order 
of 2 to 3 volts and the applied sinusoidal potential should therefore 
be of the order of 15 volts. It should be remembered that there 

^o WW 

Us 

W %woo 

.§^ 330 
■5: ^ 
11 -^ 

05 v - ago 

MO 250 500 1000 2500 5000 M000 
Natural Reiaxation Frequency 
(cycles per second) 

Fig. 172. The degree of stabilization of Builder's circuit 
in the absence of a control potential 



will be a potential drop in the phase-shifting network and this must be 
taken into account when designing the tuned transformer T 2 . 

The saw-tooth wave-form (c) is applied to the grid of the valve F 2 , 
and this is converted to a sinusoidal wave-form by the transformer 
which is tuned to the required fundamental frequency of the saw-tooth. 
If the period of the saw-tooth wave-form changes, the output from the 
transformer is greatly reduced and a phase-shift is introduced which, 
when the circuit is correctly adjusted, tends to restore the time base 
period to the correct value as determined by the output from the 
transformer T z which is tuned to the required frequency. As a result 
of this stabilization, the natural frequency of the time base can be 
varied considerably without producing a very appreciable change of 
the output frequency. This is shown in Fig. 1 72, which depicts the 
degree of stabilization when the selector circuit L X C^ has an effective 
magnification factor Q=6. It will be seen that, in this case, the output 
varies from 983 to 1,009 cycles per second for a change of the natural 
frequency of the time base portion of the circuit from 200 to 5,000 
cycles. Since the change in the natural frequency due to temperature 
and other changes in the components is unlikely to exceed 10 per cent, 



248 



TIME BASES 



the degree of stabilization is very high. The natural frequency of the 
time base may conveniently be altered by adjustment of the bias 
resistance R 2 , 

When the high-frequency control potential is applied to the primary 
winding of the transformer T lt the half-sinusoid shown at (b) in Fig. 171 
is modified as shown at (d), and the instants of commencement and 
cessation of the saw-tooth wave-form are then rigidly controlled 
by the modified wave-form. Since the tuned circuit L 2 C 4 has been 
adjusted to the required sub-harmonic frequency, the control potential 
wave-form of the high-frequency oscillator will be superimposed on 
the lower frequency with a constant phase relationship. 



OH.T-h 



Control 
Potential 
o 




Oat 



OH.T- 



Fig. 173. Builder's single- valve frequency dividing circuit 

Builder has developed another circuit employing only one valve 
which is, however, considerably more complex in operation since the 
various functions of the circuit are not separately performed as in the 
circuit of Fig. 170. The single- valve circuit is shown, for comparison 
in Fig. 173. 

Frequency Division from a Pulse 
Clayden's Frequency Divider 

Clayden* has developed a frequency divider for the division of 
pulses in which the resultant output pulse is almost synchronous with 
the leading edge of an incoming pulse. The circuit is shown in Fig. 174 
In this circuit, the valve V t has its anode current switched on by the 
arrival of a positive pulse at the grid so that the cathode of the diode V 
is driven negative and an increment of charge passes into the condenser 
C 4 . Since the anode current of V x has a high value, the stray capaci- 

19X20,' p. la i9 d o! n '* "^ ImprOVed Puke Fre <l uenc y Divider, » Electronic Engineering, 



USE OF TIME BASE FOR FREQUENCY DIVISION 249 

tances associated with the circuit are very rapidly charged and do not 
greatly delay the appearance of the charge in the condenser C 4 . The 
wave-front of the increase of potential across C 4 is made more steep 
by the presence of the inductive resistance R 7 which creates an over- 
shoot. This overshoot eventually causes the diode F 4 to conduct when 
a sufficient number of increments of charge have been passed into C 4 . 
The grid of the valve V 6 is therefore driven in a negative direction and 




Fig. 174. Clayden's frequency divider 



V 7 commences to pass anode current. The presence of the common 
cathode load resistance R l3 , R u provides positive feed-back and the 
resultant cumulative action completely cuts off the anode current in V 6 . 
The valves V % and V 7 form a multivibrator and the output may be taken 
from either or both of the anode circuits. The negative 300 volt rail is 
not at earth potential. 

When V 7 conducts, the potential of its cathode is raised and this 
causes the diode V & to conduct and to remove the charge in the con- 
denser C 4 , so that the potential of its upper plate is made equal to that 
at the cathode of the diode V 2 . This permits anode current to flow once 
more in the anode circuit of V 6 and the cumulative action reverses. 
The width of the output pulse is determined by the time constant of the 
network C 7 (R^-hRn) coupling F 6 to V-. 

The dividing ratio is controlled by the adjustment of the potentio- 
meter R-. 



250 TIME BASES 

The Cathode-controlled Sanatron Frequency Divider 

The circuit of the cathode-controlled sanatron* frequency divider due 
to Williams and Moodyf is shown in Fig. 175, This circuit is used to 
divide the repetition frequency of a series of pulses fed to the control 
terminals and not to reduce the frequency of a sinusoidal source as in 
Builder's circuits. The number of triggering pulses which are counted 
is determined by the duration of the run-down and fly-back in relation 



Ri 



f* — r-WMrt- \ 




tja 



■200V 




-O+330V 



^^UUV j— • it- 



17 



-o Output 



■GOV 
O-J50V 
Fig. 175. The cathode-controlled sanatron frequency divider 

to the time between consecutive pulses. This number is therefore 
determined by the potential at the anode of V 2 at the commencement 
of the run-down and, hence, by the setting of the potentiometer R 2 . 
In order to prevent those triggering pulses, after the first, which arrive 
before the circuit has returned to the stable condition, from affecting 
the circuit, it is necessary to apply the control pulses to the cathode 
of V lt and these pulses should have an amplitude of about —20 volts. 
The necessity for applying the pulses to the cathode of F T will be 
obvious when it is realized that, from the commencement of the run- 
down until the circuit returns to the stable regime, the diode V l is 

• Since the principle of operation of the sanatron has been described in detail in 
Chapter IX, it will not be repeated here. The santrig circuit to be described later 
in the present chapter is basically the same as the sanatron. These circuits are inserted 
to this chapter in order to provide easy reference to frequency dividers and counters 
in the same chapter although, circui tally, they belong to the Miller-capacitance croup 
of circuits. 

t F. C. Williams and N. F. Moody: British Patent Application No. 16043/44 
F. C. Williams and N. F. Moody: "Ranging Circuits, Linear Time Base Generators 
and Associated Circuits," Jotirn. I.E.E., 1946, 93, Part IIIA, p. 1188 



USE OF TIME BASE FOR FREQUENCY DIVISION 251 

cut off. When the control pulse arrives, the anode, grid and cathode 
of V 2 all fall in potential together. The cathode potential falls to a 
value, determined by the resistances R 5 and R 7 , in the neighbourhood 
of —23 volts and the grid potential falls to a greater extent so that 
anode current is nearly cut off. This part of the operation of the 
circuit has been described since it differs from that of Fig. 133 on 
page 187. 



I / \ { \ f "* Trigger Pulses 
\J \J \J to be dtytded 



Potential 
at Anode 
ofVi 




r4 

1 K Grid Base of U ? 
Fig. 176. Wave-forms pertaining to the circuit of Fig. 175 

A square wave-form is obtainable from the anode or screen grid of 
the valve F 4 , and the repetition frequency of this wave-form is a 
sub multiple of that of the triggering pulses. The precise value 
of the submultiple is determinable by adjustment of the potentio- 
meter R 2 . 

The wave-forms appropriate to this circuit are shown in Fig. 1 76, 



The Suppressor Grid-controlled Santrig Frequency Divider 

For the santrig frequency divider, the circuit of which is shown in 
Fig. 177, two sets of triggering pulses are employed. One of these 
(initial) is employed to trigger the circuit while the other (final), which 
is synchronized with the first and which has a repetition frequency 
which is an integral multiple thereof, is used to control the instant 
at which the circuit commences to return to the stable regime. 
The object of this arrangement is to ensure that the return to the stable 
condition commences at a precisely predetermined time interval after 



2S2 



TIME BASES 



the commencement of the run-down. As a result of this arrangement, 
the duration of the square wave-form appearing at the anode of V- can 
be precisely fixed. The achievement of precision in timing of the return 
to stability is due to the fact that, at the end of the run-down, the 
circuit resides in a semi-stable condition during which the grid of the 
valve V h is at a known potential which is not affected by any adjustments 
made to alter the rate or duration of the run-down. The re-triggering 
of the circuit by means of the final control pulse therefore occurs always 
at the same point on the leading edge of the pulse. 



v+JOQV 



-O +200 V 




o Output 
-o Output 



Control £R 2 ?^ 

Potential ' 
{Initial} 



o +20 V 

"GOV 

Control 
Potential (Final J 
°-JOOV 



Fig. 177. The suppressor grid -controlled santrig frequency divider 

The initial control pulses, which are negative pulses of approximately 
20 volts amplitude, are applied to the cathode of the diode V l while 
the final (high repetition frequency) pulses, which are positive pulses 
of approximately 10 volts amplitude, are applied to the anode of the 
diode V^ 

During the stable regime, the valve V 6 takes grid current via the 
resistance R l 7 , which is fed from a +20 volts supply rail and the potential 
at the anode of this valve is therefore low, as is that of the suppressor 
grid of V 2t so that the anode current in V 2 is cut off. The control grid 
of V 2 is held at, or just above, zero potential by the diode F 3 , while the 
current which flows through the resistance R 12 also passes through the 
diode. The cathode of V 2 follows that of the grid and resides at a 
potential just above zero volts. At the same time, the anode of the 
diode Vi remains at a potential of —10 volts obtained from the resis- 
tances R 7 > R s and i? 9 across which there is, during the stable regime, 
a potential of 300 volts. 



USE OF TIME BASE FOR FREQUENCY DIVISION 253 

When the initial triggering control pulse arrives at the cathode of 
the diode V l} the potential of the anode of V 2 necessarily falls, and 
tins causes a fall of grid potential of the valve V 5 , via the condenser 
C 5 and resistance R u , which reduces the anode current in V h and raises 
the potential of the suppressor grid of F 2 . If desired, a further diode 
may be connected between the suppressor grid of V 2 and the zero 
potential supply rail in order to prevent the suppressor grid from being 
driven positive with respect to this supply rail. In the absence of this 
diode, the suppressor grid is driven to a potential of about +5 volts, 
which is determined by the values of the resistances R 15 and R ia . 
Anode current therefore flows in the valve V 2 and the run-down com- 
mences while, at the same time, the grid of V 5 is driven beyond the 
cut-off potential. 

The fall in the anode potential of V 2 drives its grid negative, via the 
condenser C 4 , until it is almost at the cut-off potential. The potential 
at the anode of V 2 falls by a greater amount than that of the grid and 
cathode owing to the presence of the resistance R& which lies between 
the anode of V 2 and the condenser C 4 . 

The fall in the potential at the cathode of V 2 results in a fall at the 
anode of the diode F 4 , but the grid of F 5 , and with it the cathode of 
7 4 , also suffer a fall in potential via C 5 and i? 14 , and the final control 
pulses can therefore have no effect during this part of the cycle of 
operations. 

At the end of the run-down, the anode, cathode and grid of V 5 all 
rise in potential and the anode of F 4 returns to its stable-condition 
potential of —10 volts and carries its cathode and the grid of V z with 
it. Thus, the grid of V 5 is held at —10 volts until the time constant C 5 
{R u -\-Rn) permits the condenser C 5 to become charged. This 
potential, —10 volts, is, of course, sufficient to hold V- a in the cut-off 
condition. Because of this, the valve V 2 continues to pass anode current 
and, as the feed-back across the condenser C 4 has ceased, the anode 
potential of V 2 remains constant. A semi-stable regime is thus engen- 
dered and this continues until the arrival of the next final triggering 
control pulse at the anode of F 4 . When this occurs, the anode of V 4 
is driven positive followed by its cathode, the pulse being developed 
across the resistance R u . As a result, the potential of the grid of V s 
rises rapidly, its anode potential falls, the potential of the suppressor 
grid of V 2 falls and the anode current of V % is once more cut off so 
that the circuit returns to the stable regime. 

As already stated, the output may be taken from the anode or screen 
grid of the valve V 5 and the precision obtainable in the duration of the 
output square wave-form is equal to the time interval between the 
initial pulse and the selected final pulse to within ±0-3 microseconds 



254 



TIME BASES 



or less, without regard to the number of final pulses which occur before 
that selected to initiate the return to the stable condition. The degree 
of precision is increased as the sharpness of the final pulses is increased. 

The wave-forms which are appropriate to this circuit are shown 
in Fig. 178. 

The following values may be employed for the various component 
parts of the circuit: 



R 2 

*c 

-^11 
R u 

"^16 



10,000 

50,000 

5,000 

12,000 

200,000 

10,000 

2,000 

10,000 

30,000 

470 
470 

6 

3,000 

10,000 

47,000 

30,000 



ohms 

ohms 

ohms 

ohms 

ohms 

ohms 

ohms 

ohms 

ohms 

ohms 

ohms 

megohms 

ohms 

ohms 

ohms 

ohms 



R 17 

^18 



5 megohms 
470 ohms 
R 19 15,000 ohms 
R ZQ 10,000 ohms 
Cj 0*001 up. 

C 2 o*i -jiF. 

C 3 0-5 (iF. 

C 4 000 1 [iF. 

C 5 o-ooi ixF. 

C 6 o-ooi [j.F. 

C 7 o-ooi pF. 

V x VR92 
V 2 VR91 
V s VR92 
V 4 VR92 
V s VR91. 



The circuit described has suppressor grid-control of the valve V 2 . 

Scalariform Frequency Divider or Time Base 

The circuit of a scalariform frequency divider or time base, employ- 
ing Germanium crystal detectors, is shown in Fig. 179, in which the 
key represents a discharging device, usually of the trigger type. 

The application of a potential of square wave-form to the input 
terminals results in an increment of charge appearing in both conden- 
sers during one half-cycle. During the following half-cycle the incre- 
ment of charge already acquired by the condenser C t is removed. C 2 
thus acquires a succession of similar charges during alternate half- 
cycles and these result in a scalariform output potential. The discharge 
circuit removes the charge in C 2 when the latter reaches a predetermined 
potential and each time this occurs the output potential falls substanti- 
ally to zero. 

The use of Germanium crystals is advisable because of their very low 
value of forward resistance. It is, of course, essential that the output 
potential be not greater than a small fraction of the input potential 



USE OF TIME BASE FOR FREQUENCY DIVISION 255 

since, otherwise, the scalariform output becomes exponential instead of 
linear. 

This circuit has the very great advantage that the frequency of the 
input square-wave may be varied over a fairly large range of frequency 




'A—JZJJ 



Anode V s 



_ZQ53CZK! 



Final Trigger 



y-+20QV. 



— +30 K 



E3ZZOC 



Fig. 178. Wave-forms pertaining to the circuit of Fig. 177 



II 



Square 

Wave - form 

Input 



Ul 



Bz 



f »T f 



Output 



Fig. 179. Scalariform frequency divider or time base employing 
Germanium crystal rectifiers 



without any change in the component values or in the frequency 
division ratio. The frequency division ratio is determined by the values 
of the ratios: 

C x , e 

— — ^— and — . 

C\ t\ E 



256 



TIME BASES 



Thus, if the peak-to-peak input potential is E and the trigger potential 
is e, the frequency division ratio R is 

R= -_ eC l , 

E{C X + C 2 ) 

It will be realized that particularly where the division ratio is a high 
one, there is sometimes a possibility that the frequency divider may 
operate on the signal which precedes the correct one. Scantlebury* has 
evolved a device to prevent this. The period immediately before the 
arrival of the correct pulse is called the vulnerable period since it is 
possible for some random disturbance to raise the amplitude of the pulse, 
or signal, preceding the correct one, to a value sufficient to operate the 
circuit. In his circuits, Scantlebury reduces the duration of the vul- 
nerable period, continuously measures it and continuously and auto- 
matically adjusts the time of commencement of this period. Thus, if 
the correct operating signal is delayed for a few repetitions, the com- 
mencement of the vulnerable period also becomes delayed. This device 
is, of course, applicable to many time bases and frequency dividers, but 
it is more applicable to valve circuits. 

Moss's time base as shown in Fig. 27, page 45, may also be used as 
a frequency divider. 



COUNTING 

Countingf involves the integration of a number of random input 
pulses whose quality may be specified as lying within certain dimensional 
limits but with wide temporal tolerances. The input pulses often require 
to be amplitude limited. Because of these wide temporal tolerances, 
a high degree of stability is required during the stable regime. There- 
fore only those frequency dividers having high stability can be em- 
ployed as counters even if the input wave-form is suitable. On the 
other hand, a counter may be used as a frequency divider provided that 
the wave-form whose frequency is to be divided is such as to make it 
capable of operating the particular counter to be employed. Counting 
is often carried out by devices which are not time bases, but these will 
not be considered here. 

A high degree of circuital stability is essential in a counting circuit 
and valuable in a frequency divider. The period for which stability 
must exist for a frequency divider is precalculable. Although for a 

*G. S. P. Scantlebury; British Patents 595307 and 595345. 

t W. B. Lewis: Electrical Counting, Cambridge University Press; I. E. Grosdoff: 
"Electronic Counters," R.C.A. Review, 1946, 7, p. 438; C. B. Leslie: "Megacycle 
Stepping Counter," Proc. LR.E., 1948, 36, p. ^30; E. L. C. White: British Parent 
47I73I- 



TJSE OF TIME BASE FOR COUNTING 



257 



counter, this period is theoretically infinite, in practice, a limit may be 
imposed when considering the data on which the design of the circuit 
is to be based. 

Benjamin's Blocking Oscillator Counter 

The circuit of Fig. 180* is stable and will operate with a high ratio 
of input to output pulses. Negative trigger pulses limited to a constant 

-OH.T.+ 




Input 



0//7:- 



Fig. 180. Benjamin's blocking oscillator counting circuit 




Input 

m 



Cathode 
ofV s 



Output 



Time 



Fig. 181. Wave-forms pertaining to the circuit of Fig. 180 

amplitude e volts greater than the cathode-grid potential x of the 
cathode follower valve V z are required to operate the circuit. The 
input wave-form is shown at (a) in Fig. 181, from which it will be seen 
that the input pulses may be random in time and duration, but the 
amplitude must, as already stated, be limited to a constant value. 

* R. Benjamin: "Blocking Oscillators," >»r«, I.E.E., 1946, 93, Part IIIA, p. 1159. 
19 



25» 



TIME BASES 



When the circuit is put into operation, a potential appears across the 
resistance R 5 which causes the valve V b to conduct; the cathode current 
of V 5 then charges the condenser C 2 via the diode V v When the first 
incoming pulse arrives, C 2 is partly discharged via V 2 , whose cathode 
is made negative by the incoming signal. The potential across C. 
therefore falls by x+e volts. At the end of the pulse, the cathode 
potential of V z rises and its anode current is cut off. At the same time 
V 1 commences to conduct and the grid potential of V z becomes 

€ 'C~+~C~ because the condensers C x and C 2 form a potentiometer while 

V x is conducting. At this stage, the potential of the grid of V m and 
hence of the cathode of F 5 , is stable, but the potential thereof is reduced 

by e ' c +C Which is a PP roximate] y equal to e volts. The initial 
starting conditions can be adjusted by a variation of the slider of the 
potentiometer i? 4 ; the potential tapped off determines the maximum 
positive excursion of the anode of F 4 and, hence, of the cathode of 

Each input pulse repeats the procedure until the grid of F 3 and the 
cathode of F 5 are brought sufficiently low in potential for F 6 to conduct 
The blocking oscillator V 5 will then operate for one cycle and its 
cathode current will recharge the condenser C 2 via the valve V x while 
the diode F 4 , as already stated, determines the starting conditions 
The counting ratio is thus set by controlling the position of the slider 
of R 4 or by altering the amplitude of the incoming pulses. 

The following values may be employed for the components: 



-#i 5-6 megohms 

R 2 5>°oo ohms 
C x 70 p.u.F, 



C 2 



500 |*|*F. 

0-1 up. 
o-i [iF. 



i 



APPENDIX I 
THE CATHODE RAY TUBE 

Cathode rays were first investigated in 1859 by Plucker, who gave 
them this name presumably because he found that they emanate from 
the cathode. In 1890 Johnstone Stoney discovered that the already 
known fluorescence produced when cathode rays strike the glass 
walls of the vessel in which they are generated is caused by bombard- 
ment of the walls by particles which he called electrons.* Cathode 
rays consist of large quantities of these electrons, which are minute 
particles of negative charge, all travelling in substantially the same 
direction. 

The first useful applications of the cathode ray tube were made 
independently by Braunj- and Kaufmann £ in Germany and by 
Thomson § in England. Braun used the tube for tracing curves and 
Thomson and Kaufmann used it to measure the ratio efm. These 
early tubes were very simple structures which required a potential of 
50,000 to 100,000 volts for their operation, 

A great deal of early development work on the cathode ray tube was 
carried out by Fleming|| and others at the end of the nineteenth and 
the commencement of the twentieth centuries. 

Various improvements were made by numbers of workers both in 

* From the Greek word tf\eKrpoi> t meaning amber. Amber has long been known to 
have electrical properties. These properties are said to have been first discovered when 
amber was rubbed with other material by Thales about 600 B.C., although he did not 
understand the effect. Pliny (a.d. 70) also knew of the effect. About the year 1600 
Gilbert announced that materials other than amber also exhibited electrical properties. 

f F. Braun: "tJber ein Verfahren zur Demonstration und zum Studium des 
zeitlichcn Verlaufes variabler Strome" ("On a method of demonstrating and studying 
the time duration of variable currents"), Ann. der Physik, 1807, 60, p. 552. 

% W. Kaufmann: "Die magnetische und elektrische Ablenkbarkcit der BequereU 
strahlen und die scheinbare Masse der Elektronen" ("The magnetic and electric 
deflection of Bequerel Rays and the apparent mass of the electron '*), Nach. von der 
Gesell. der Wiss. zu Gottingen, November 8th,*iooi. 

§ J. J. Thomson: "Cathode Rays," Phil. Mag., 1897, 44, p. 293; "Cathode Rays," 
Electrician, 1897, 39, p. 104; "On the Cathode Ravs," Proc. Camb. Phil. Soc, 1897, 
44, p. 243. 

I! A. Fleming: "An experiment showing the deflection of Cathode Rays by a 
Magnetic Field," Electrician, January 1st, 1897; E. Weichert: " Experimented Unter- 
suchungen uber die Geschwindigkeit und die magnetische Ablenkbarkeit der Kathoden- 
strahlen" ("Experimental researches on the velocity of cathode rays and the possi- 
bility of deflecting them magnetically"), Ann. der Physik, 1898, 69, p. 739; A. Wehnelt: 
" Empfindlichkeitssteigerung der Braunschen Rohre durch Benutzung von Kathoden- 
strahlen geringer Geschwindigkeit" ("Increasing the sensitivity of the Cathode Rav 
Tube by the Use of low velocity rays"), Phys. Zeits., 1905, 6, p. 732; A. Campbell"- 
Swinton: Letter to Nature, 1908, 78, p. 151; D. Roschansky: "Ober den Einfluss 
des Funkens auf die oszillatorische Kondensatorentladung" ("On the influence of 
the spark on the oscillatory condenser discharge"), Ann. der Physik, 191 1, 36, p. 281 ; 
A. Campbell-Swinton: Presidential Address to the Rdntgen Society, Journ. Ront. 
Soc, 1912, 8, p. 1. 

259 



260 



TIME BASES 



Europe and America, but it was not until 1922, when Van der Bijl 
and Johnson* developed the first practicable low-voltage gas-focused tube 
that the new instrument was able to play a really useful part in scientific 
investigation and development. The history of the cathode ray tubef 
is an interesting one, and is worthy of attention. 

The modern high-vacuum electrostatic cathode ray tube consists of 
a source of electrons, a control electrode, a focusing system, a deflecting 
system and a viewing screen, all of which are enclosed in a glass 
bulb. The cathode is indirectly heated and has a small area of emission ; 
it is surrounded by a Wehnelt cylinder, or other electrode (generally 
called a grid), which controls the emission of electrons from the cathode. 
A fixed bias may be applied to the grid and a varying potential may be 
superimposed in order to vary the instantaneous brightness along the 
trace, in much the same way as the anode current of a valve may be 
varied by a potential applied to the grid. 

In a gas-focused tube there is an object lens produced by the potential 
gradients between the anode, cathode and grid, and the beam so formed, 
which contains an image of the cathode, is brought to a focus by the 
gas which becomes ionized. The positive ions which are formed by 
collision between the electrons and the gas molecules form a core down 
the centre of the beam and, since they are positively charged, they 
attract the electrons inwards, thus bringing the beam to a focus at the 
screen. The degree of ionization and, hence, the point in space at 
which focusing occurs is controlled by altering the grid potential and 
so adjusting the emission. 

Two great disadvantages of the gas-focused tube have caused it to 
be superseded by the high-vacuum tube. These disadvantages are loss 
of focus at high writing speeds and origin distortion,! which is a reduc- 
tion of deflection sensitivity occurring for a few volts on either side of 
zero deflection potential. 

The focusing means employed in a high-vacuum tube differ funda- 
mentally from those employed in a gas-focused tube. The high- 
vacuum tube employs two lenses, an object lens and a final lens. The 
object lens is formed by the potential gradients between the first anode 
(nearest the cathode), the grid and the cathode. The final lens, which 
brings the resultant beam to a focus at the screen, is formed by the 
potential gradients between the various anodes. One of the anodes, 
generally the second anode, is called the focusing electrode and has a 

• J- B. Johnson: Journ. Amer Optical Soc. and Rev. of Set. Irnts., 1922, 6, p. 701 

44 Nos it t'Ji ,', P M u ;£ UCkle: Cat h d , e Ra ^ Tubes '" Wireless World, 1939, 
i T°V ' and l2 ' M? rch l6th - P- 242, and March 23rd. p, 275. 

len« fnr,,;^ lg n 0tt n u Com P arativ e Performance of Gas-focused and Electron- 

79 p 2T 0sciI1 °6 ra Phs at very High Frequencies," Journ. I.E.E., 1936, 



APPENDIX I 



261 



variable potential applied to it for the purpose of determining the point 
in space at which the focus occurs. The entire system for forming the 
electrons into a beam and for bringing them to a focus at the screen is 
known as an electron -optical lens system* or an electron gun. 

It is of the greatest importance that all the lens and beam-forming 
component parts of the cathode ray tube, together with the screen 
surface, be assembled with the most scrupulous care in order to pre- 
serve geometrical accuracy since, otherwise, the spot cannot be circular 
and/or will not remain circular when it is deflected. All the lens-forming 
apertures must be correctly dimensioned and well polished in order to 
provide a fine spot without aberration. 

The anodes, which are a part of the lens system, are in the form of 
discs pierced with central holes or of cylinders or of a combination of 
discs and cylinders. In general, the potentials applied to these anodes are 
fixed in value except for that applied to one of the anodes which is made 
adjustable as a fine control of the focus. When the appropriate poten- 
tials are applied to the different portions of the focusing system, the 
electrons emitted from the cathode pass through the discs or cylinders 
and are there formed into a beam which is brought to a focus 
at the screen. This screen consists of a quantity of fluorescent 
crystalsf adhering to the inside of the glass wall of the tube at the end 
opposite to the cathode. When the beam strikes the fluorescent 
screen, light is emitted at the point of impact and electrons leave this 
point by secondary emission, whence they return to the final anode, 
thus completing the electrical circuit. A point of light thus makes 
visible the position of the end of the cathode ray beam where it 
bombards the screen. 

Since the beam is composed of particles (electrons) moving with a 
high velocity, it has practically no inertia, and can therefore be swung 
to and fro many millions of times a second. 

The Production of a Double Beam 

Fig. 182 shows Fleming-Williams' arrangement for producing a 

• E, B ruche and O. Scherzer: Geometrische Elektronenopfik (Geometrical Electron 
Lenses), Verlag Julius Springer, Berlin, 1934; L. H. Bedford: "A note on Davisson's 
Electron Lens Formulae," Proc. Physical Society, 1934, 46, p. 882; V. K. Zworykln: 
"On Electron Optics," Journal of the Franklin Institute, 1933, p. 535 ; M. von Ardenne : 
Television Reception, Chapman and Hall, London, 1936; L. M. Myers: Television 
Optics, Pitman, London. 1936; O. Klemperer: Electron Optics, Cambridge University 
Press, 1939; H. Moss: "The Electron Gun of the Cathode Ray Tube," jfouni. Brit. 
I.R.E., 1946 (Part I), 5, p. 1 ; (Part 2), 6, p. 99- 

t L. Levy and D. West: "Fluorescent Screens for Cathode Ray Tubes for 
Television and other Purposes," Journ. I.E.E., 1936, 79, p. 1 1 ; R. B. Head: "The 
Luminescence of Simple Oxide Phosphors," Electronic Engineering, 1948. 20, p. 219; 
H. W. Leverenz: "Luminescence and Tenebrescence, " R.C.A. Review, 1946, 7, 
p. 199. 



262 



TIME BASES 



double beam" m one oscillograph tube. A splitter plate, which is 
immersed in the beam, and which divides it into two separate section 
can clearly be seen, together with two "bucking" wires B to which 
potenfals are applied in order to cancel out mutual deflectional inter- 
Sn e ,r a , J I tW ° bcam 1 S - With this tube, the two beams are simul- 
TZvP V^ X 3XiS but ^ are «I»»«««y ""trailed 

unbalanled'T T ** Y ^'^ P 0tentials are n^essnrily 

unbalanced, because only one plate is available for each beam, it 

becomes necessary to employ an anti-trapezium construction for the 
1 J™ ."^ruction, which is also due to Fleming-Williams, is 
shown at J in F lgs , ,82 and 187 and is discussed on page 276 

nil, If 7 ^ e ' eCtr0de aSSembl 5' mountcd o" °™ of several 
possible types of cathode ray tube base. 

The production of a double beam is much better accomplished by 
means of Flemmg-WUW double-beam cathode ray tube than by 

anneTri "" ^ "^ ***• ™ ^ former case ' the two images 

E n °V? COmC ! dent "V*"? This is not *° with the electronic 

timet,? 4 ', 'l P ° SSlble fOT CTentS WhiCn are n0t C0inciden t in 

time to be assumed to be so. 

The M^chanL a r ° f a Cath ° de Ray Tube as - bating 

It is apparent that the high-vacuum cathode ray tube forms an ideal 
"Seating or delineating mechanism with the following cha^cteristic": 

{a) l ntt T bk °/ ° Pe n ati ° n at mUCh hi ^ her frequencies than any 
other form of oscillograph and, moreover, the sensitivity is 
independent of frequency at least up to about 100 megacycles 
per second. ° J 

(b) It is the only form of oscillograph in which the light intensity can 
. be modulated at high frequencies. * 

(c) The load due to the oscillograph on the source to be measured is 
generally negligible so that distortion from this cause need not be 
considered at least for frequencies up to about 500 kilocycles per 
second For higher frequencies the stray capacitances must 
certainly be considered although they can often be made to form 
part of a tuned circuit. For still higher frequencies the load due 
to the presence of the stray capacitance may involve the inclusion 
of a power amplifier in order that the input to the cathode ray 
tube deflector system may be provided from a low impedance 
source. At very high frequencies (order of io» cycles per second) 
the transit time of the electrons forming the beam is such that 

^ „£^K <*■»* «* OsciHograph,'* 




Fig. 182. Fleming-Williams' double beam cathode ray tube (A. C. Castor, I. id. 

London). The shield has been partially cut away to permit the splitter plate to be 

seen. This is also an anti-trapezium distortion tube 



[To face page 262 



APPENDIX I 



263 



almost a complete cycle of the deflecting potential can occur 
while they are in the act of passing through the length of one 
pair of plates. As a result the electrons are deflected from side to 
side between the plates and may emerge with little or no deflec- 
tion. This is known as the Hollmann* effect and the remedy is to 
reduce the length of the deflecting plates. Special cathode ray 
tubes have been designed by Hollmann to overcome the effect 
(see page 278). 
(4) It provides its own light source (unlike a mechanical oscillograph). 

The cathode ray tube is one of the most versatile of tools and with 
its aid an enormous number of different observations and measure- 
ments can be made both with and without the use of a time base. This 
versatility is due to the fact that the tube functions as a voltmeter or 
ammeter which indicates instantaneous values of the quantities con- 
cerned. The field of investigation which may be undertaken with the 
cathode ray tube is enormously widened by the production of potentials 
and/or currents which are proportional at every instant to mechanical 
or physical quantities. Of course, two potentials or two currents or a 
potential and a current may be compared. The tube is, of course, put 
to many objective purposes such as television, the transmission of 
handwriting, etc., which are not "means to an end" as is the case with 
observations and measurements. There are many developments from 
the cathode ray tube, some of which include a cathode ray beam- 
forming device together with a photo-electric device. These are used 
for the transmission of scenes for television. Other devices employing 
cathode ray beams are used for the generation of electrical energy at 
extremely high frequencies. Cathode ray beams have also been em- 
ployed as switching devices, notably for multiplex signalling over a 
single line on a time-sharing basis. Hollmann has developed a special 
tube for three-phase investigations and has applied this to electro- 
cardiography, in which potentials from each wrist and one ankle of a 
patient are examined and compared. 

OPERATING POTENTIALS FOR THE CATHODE RAY TUBE 
In order to be able to use the cathode ray tube it is necessary to 

know the deflectional sensitivity of the tube and the ratio of the 

potentials to be applied to the various electrodes. 

Most modern high-vacuum cathode ray tubes require about 3,000 

to 8,000 volts on the final anode (nearest the screen) for optimum 

results, but some tubes are made which work well with 1,500 volts or 

*H. E, Hollmann: "Die Braunsche Rohre bei sehr hohen Frequenzen" ("The 
Cathode Ray Tube at very high frequencies"), Zeit. Hachfrequemtechnik, 1932, 40, 
p. 127. 



264 



TIME BASES 



less. The focusing electrode normally requires a potential of about 
one-sixth to one-third that of the final anode. If the second anode 
is the focusing electrode, the first anode is supplied either with a 
potential equal to about one-tenth of the potential on the final anode 
or is directly connected thereto, sometimes internally. The modulator 
is usually supplied with a potential of the order of one-thirtieth to 
one-hundredth of final anode potential, usually negative. The appro- 
priate potentials required for any particular design of tube are always 
quoted by the manufacturer. When referring to the potential of 
the anode or of any other electrode of a cathode ray tube or valve the 
potential difference between the electrode concerned and the cathode 
is implicitly mtended unless the potential is quoted with respect to 
earth or to some other specified point in the circuit. There is one 
exception to this rule as follows: when the potential under discussion 
is applied to a deflector plate, the potential intended is that appearing 
between the deflector plate concerned and the final anode of the 
cathode ray tube. 

Normally, one of the time base supply rails is connected to the final 
anode of the tube but, occasionally, the anode is connected to another 
part of the time base circuit. This occurs particularly where no con- 
densers are connected between the time base and the deflector plates 
and the changed connection is therefore required to ensure the correct 
mean potential conditions (described on page 278) between the deflector 
plates and the final anode of the cathode ray tube. 

The current which a cathode ray tube takes from the power supply 
apart from that necessary to heat the cathode, rarely exceeds half a 
milhampere. For this reason, resistance-capacitance smoothing is a 
matter of practical economics since, as the current is small, a high 
degree of smoothing can be obtained without the necessity of wasting 
an appreciable percentage of the high tension supply potential. It is 
important to ensure that no ripple remains beyond the smoothing 
system. This may be arranged by connecting condensers across all 
points of the supply which are connected to the tube and by inserting 
a resistance between the smoothing system and the tube. This is 
shown in Fig. 183, which depicts a voltage-doubler rectifying system 

As an alternative scheme, when it is desired to obtain a rectified 
output potential which is higher than that available across the secondary 
terminals of the transformer, Cockcroft and Walton's* method may 
be employed. A circuit for obtaining potential multiplication by a 

* J;,*£ Cockcroft and E. T. S. Walton: "Experiments with High Velocity Positive 
Ions, • Proa Royal Society (A), 1932, 136, p. 619; E. B. MoulHn: ''External ChaT 
actenstics of a Diode Rectifier," Journ, LEE ioi7 80 d «*• C U ri«. 



r 



r 



APPENDIX I 



265 



factor of 2 is shown in Fig. 184 (a). In this circuit, the potential between 
the junction of the valves V 1 and V z and earth varies between — zE 
and zero, where E is the peak potential across the secondary winding 
of the transformer. The potential across the condenser C 1 is approxi- 
mately E volts, the condenser being charged through V 1 when the 





focus , ,_ I 




Rz%^T. 



3^J_A_**«M»«y 



Control' 
Fig. 183. Voltage-doubler power supply circuit for a cathode ray tube 




Fig. 184. («) Cockcroft and Walton's voltage-doubler power supply circuit 
(b) voltage-quadrupler power supply circuit 

upper end of the secondary winding is positive. When this end of the 
winding becomes negative, the condenser C 2 is charged through F 2 
and the potential across the condenser C 1 is added to that across the 
secondary winding of the transformer. The potential of the source 
which impresses a charge upon the condenser C 2 is thus zE volts 
approximately, and the circuit therefore behaves as a voltage doubler 
circuit. 

In practice, the potential across C x will be less than E volts because 



266 



TIME BASES 



TOP TEBMIHAL 
55kV 



«2SkV 



2-S5kV 



of losses in the transformer and diode and hence the potential across 
C 2 will be rather less than 2E. 

Fig. 184 (b) shows a method of quadrupling the potential of the 
transformer, and obviously the method can be extended to provide 
greater multiplying factors. Each time the potential is increased the 
available current is, of course, reduced by the same factor. Cockcroft 
and Walton's circuits are of particular value where the potential is to 
be raised by a factor greater than 2. 

Walker* has developed a system of 
rectifiers, with smoothing condensers, 
based on Cockroft and Walton's 
method, which may be employed for 
the production of extra-high-potential, 
direct current supplies, for use with 
cathode ray tubes. The rectifiers and 
condensers may be purchased in the 
form of a unit which may be tapped 
across an existing transformer supply- 
ing high tension at, say, 400 volts for 
other purposes. The arrangement is 
shown in Fig. 185. 

It will be noticed that the trans- 
former has a centre tap on the H.T. 
winding to enable it to be used with 
the full-wave rectifier valve. This has 
the effect of adding an alternating 
component to the steady E.H.T. 
potential but this component is re- 
moved by the addition of two extra 
rectifiers, one at each end of the 
E.H.T. chain. 
Extra-high-potential supplies may also be obtained from the fly- 
back potential generated across the deflection coils, or deflection trans- 
former, and this method has the advantage that if the time base ceases 
to function, the H.T. supply to the cathode-ray tube fails and, as a 
result, it is impossible for the tube screen to be burnt by a stationary 
spot. 

Schadef has developed a method of obtaining the high potential 
required for operating the cathode ray tube from the unidirectional pulses 
obtained from the line scanning transformer during the fly-back period. 

•AH. B. Walker: "Television E.H.T. Supply," Wireless World, iy 4 8, 54, pp. 120 
and 169. bee also "A New Rectifier E.H.T. Unit/' Electronic Engineering, 1948, 20, 
p. 144. 

t O. H. Schade: British Patent 613053 ; W. R. Koch: British Patent 618624. 




Fig. 185. Walkers' voltage-multiplier 

for use as a cathode ray tube power 

unit 



APPENDIX I 



267 



The circuit of Schade's device is shown in Fig. 186. The fly -back 
pulse is fed to the transformer 7\ where its potential is stepped up 
and it is then applied to the diode V t in series with the condenser C x . 
During the pulse, the condenser becomes charged and, during the next 
line scan, it discharges into C 2 via the resistance R v When the next 
pulse appears, the right-hand plates of C t and C 2 are both positive, so 
that the potential across the two condensers in series is considerably less 
than that of the pulse which has been used to provide the charge. When 
the next pulse arrives, therefore, the condensers C\ and C 3 are both 
charged in the same sense and C 2 is discharged. As a result, the charge 
across C v and C 3 in series is greater than the potential of the fly-back 




Fig. 186. Circuit of Schade's voltage-multipher 

pulse as it appears across the secondary winding of the transformer T v 
The resistance R x and condenser C 4 are employed to remove ripple. 

Obviously, more stages may be employed to increase the resulting 
potential. Only a small current may be drawn from the system as the 
regulation is not very good, but it is quite capable of supplying the 
necessary current for operating a cathode ray tube. 

Another method of obtaining the high potential is to generate a high 
frequency* of the order of t megacycle using a power oscillator valve, 
to increase the potential with the aid of a tuned transformer and to 
rectify the output for application to the cathode ray tube. 

Since the optimum focus is obtained only when the potentials applied 
to the focus electrodes of the cathode ray tube have a certain ratio, the 
focus may be upset when the beam current is modulated by the applica- 
tion of a variable potential difference between the grid and cathode of 
the tube. This occurs, because, although a change of beam current does 
not appreciably alter the value of the current flowing to A 2 (see Fig. 187) 
it does change that to A 1 and A z% so that the focus changes slightly. 

Blumlein and Spencerf have developed a circuit to overcome this 

*K. C. MacLeod: "Radio Frequency E.H.T. Supplies," Electronic Engineering, 

1948, 20, p. 264. (Extract only — the full paper appears in the Journ. Television Society, 

1949, 5, P- aoi.) 

f A. D. Blumlein and R. E. Spencer: British Patent 480355. 



268 



TIME BASES 



effect A resistance R 7 is inserted between R a and the junction of C A 
and the two anodes in Fig. 183. The value of R 7 is made equal to 



where I z 

h 

k 



Ri~hk ( RaR » \ 
h Kk+rJ* 



R, 



is the current in A 2 , 

the current in (A t +A 3 ) r and 

the ratio of the anode potentials required to focus the 

tube when the anode currents are I 2 and / 3 . This equals 

\~r — / in the ab sence of the resistance R 7 , where 

is the resistance between the slider of * 4 and earth, and 
the resistance between the slider of i? 4 and the slider of R 6 . 
It is advisable to make the time constant 

R~C — C ( ^»^6 \ 

74 ~ c '° \K+Rj 

so that sudden changes of brightness cause the changes in the potential 
applied to the focusing anodes to occur at the same time. For tele- 
vision reception with 25 frames per second, these time constants should 
not be less than about 0-1 second. With this circuit, the lower plates of 
the condensers C 4 and C 5 are to be connected to the cathode. 

DEFLECTION METHODS 
Two methods are available for deflecting the beam: 

(a) Electrostatic. 

(b) Electromagnetic. 

In the electrostatic method, the deflecting system consists of two 
pairs of plates, arranged mutually at right angles, between which the 
beam passes. If one plate of a pair is made positive with respect to 
the other, electrons, which are negative particles, are attracted towards 
1 -and the point of fluorescence on the screen moves likewise. Fig- 187 
shows a typical cathode ray tube structure arranged for electrostatic 
deflection, with the glass bulb removed. 

In this figure the following designations are employed: 
Px 1 and Px z X x and X z deflector plates 
Pjj and Py 2 Y 1 an d Y 2 deflector plates 

A special shield which, together with the curved 
leadmg edges of the X plates, completely 
prevents trapezium distortion (see page 276). 
A-i Final anode 







Fig. 187. Cosset" single beam cathode ray tube: having electrostatic 

focusing and deflecting arrangements. This tube is also fitted with 

an anti-trapezium distortion system 

[To face page z68 



^2 

G 

C 
H 



APPENDIX I 

Focusing anode 

First anode 

Grid or modulating electrode 

Cathode 

Heater (inside the cathode). 



269 




In the electromagnetic method, the deflection of the cathode ray 
beam is brought about by causing the beam to pass through a transverse 
magnetic field. This may be applied by means of a pair of coils suitably 
disposed about the neck of the tube. Since the beam is composed of 
electrons in motion, it produces an encircling magnetic field around it 
and, hence, the external magnetic field exerts a deflecting force upon 
the beam. If the beam is coming up (through the paper) and the 
magnetic deflecting field has its North pole on the left and South on the 
right, the beam will be deflected downwards, 

For measuring purposes, as distinct from, say, television reception, 
it is usual to employ electrostatic deflection, in which case the tube 
contains two pairs of deflector plates, as already described. When, 
however, the tube is required for television, it is more normal to 
employ electromagnetic deflection applied by means of external coils. 
Television tubes also differ from general purpose tubes in that magnetic 
or electromagnetic focusing is usual and, in this case, the electrode 
system in the tube is, therefore, reduced to a cathode, a modulator grid 
and one disc, or cylinder, anode, although two anodes are often used. 

The use of magnetic focusing and deflection enables the tube to be 
reduced in length because the length of a magnetic focusing system 
is less than that of the corresponding electrostatic type. An incidental 
advantage is that since the focusing and deflecting systems lie outside 
the cathode ray tube, the cost of replacing faulty tubes is reduced. 
Furthermore, the maximum angle of deflection, without the introduc- 
tion of defocusing, is greater with magnetic or electromagnetic deflection 
than it is in the case of electrostatic deflection. The tube may thus be 
arranged horizontally for direct viewing without occupying an exces- 
sive amount of floor space. Occasionally, electromagnetic deflection is 
used in addition to electrostatic deflection for measuring purposes, but 
electromagnetic deflection is much less flexible in that special coils are 
required for each purpose. This method of deflection is, however, 
useful at power frequencies for low potential, high current measurement 
and investigation. 

DEFLECTION SENSITIVITY 
Electrostatic Deflection Sensitivity 

The electrostatic deflection amplitude obtained with a cathode ray 



270 



TIME BASES 



tube having plane parallel plates may be calculated from the following 
equation : 



Deflection 



_ EpLl 
zVd' 



1 



» 



P 



L— ~ 



J 

Fig. 188. Method of measuring the deflection 
sensitivity of a cathode ray tube 



where Ep is the peak to peak value of the deflecting potential, L the distance 
from the centre of the deflecting plates to the screen, / the length of the 

■■ deflector plates of one pair, 

V the final anode potential, 
and d the distance between 
the deflector plates of one 
pair. All these distances must 
be measured in the same 
units and the amount of the 
deflection will, naturally, also 
appear in the same units. 
Thus, the deflection is propor- 
tional to the deflecting poten- 
tial and inversely proportional 
to the final anode potential. 
The deflection is towards the 
more positive plate. 
The sensitivity of the two pairs of plates in any one tube differs if 
they are of the same size and distance apart because the effective length 
L of the beam is not the same for each pair. The calculated 
deflection may be checked by applying a measured alternating potential 
of sinusoidal wave-form to the two plates of one pair and by measuring 
the amplitude of the deflection. The potential may be applied through 
a centre-tapped transformer or through condensers and leaks as shown 
in Fig. 188. 

The maximum peak to peak value of the deflecting potential for the 
length of trace, as measured on the screen, is equal to the potential 
shown on the voltmeter, connected as in Fig. 188, multiplied by 2V2 or 
2-828. This factor is required to take account of the fact that, for a 
sinusoid, the potential shown on the meter, which is known as the RMS 

(root mean square) value, is ~ of the peak to peak value The 

2\/2 

sensitivity for the particular pair of deflector plates concerned is thus 
obtained from the following equation : 

Sensitivity (electrostatic)— — -, 

Ep 

where D is the deflection in any convenient unit, Ep the peak to peak 



APPENDIX I 



271 



potential applied (2-828 Vm) and Vm is the potential as shown on the 
meter. 

The sensitivity is usually measured in millimetres per volt, but some 
makers of cathode ray tubes quote the value of the reciprocal of sensi- 
tivity, i.e. volts per millimetre. 

It is, of course, possible to connect the potential to be used for 
sensitivity measurement to one plate only, the other end of the source 
being connected to the final anode as shown in Fig. 189. In this case, 
the potential shown on the meter multiplied by 2-828 is again the peak 



It 



tt 




Fig. 189. Alternative method of measuring 
the sensitivity of a cathode ray tube 



to peak deflecting potential Ep 
and the sensitivity is as 
before. It will be appreciated 
that the deflection produced 
by a potential of 100 volts is 
substantially the same whether 
it is connected between the 
pair of plates or between one 
plate and the final anode. In 
the former case, half the 
potential appears on each plate 
with respect to the final anode. 

It is important to note that 
any deflector plates which are 
not in use must be connected 
to the final anode as otherwise they will collect a charge which will 
often deflect the beam right off the surface of the screen, particularly 
if no leak resistance is present. 

Potentials up to 1,000 volts may, in general, be connected to the 
deflector plates without damage. The limiting factor is usually the 
insulation of the tube mounting base. 

In some cases, the deflector plates are brought to terminals at the 
sides of the cathode ray tube and it is then possible to apply considerably 
higher potentials since the insulation is improved and the leads to the 
deflector plates are widely spaced. 

When asymmetric deflection is employed, the deflection sensitivity 
is reduced as the deflector plate becomes more positive, but the effect 
is not very serious. Apart from this effect, the deflection is linear at 
all frequencies up to that at which an appreciable part of a cycle of the 
deflecting potential frequency takes place while an electron is passing 
between a deflector plate pair. This frequency normally lies at about 
10 9 cycles per second and is dependent upon the final anode potential 
and the length, along the tube, of the deflector plate. At the critical 
frequency, zero deflection occurs and the deflection then increases as 



272 



TIME BASES 



the frequency is increased. There is a series of maxima and minima 
for the deflection as the frequency continues to increase, but each suc- 
cessive maximum is reduced in amplitude. Fig. 190 shows the type of 
result obtained. This is known as the Hollmann effect (see page 262). 

Electromagnetic Deflection Sensitivity 

If a pair of coils be arranged in aiding connection about the neck of 
the cathode ray tube in such a way that the flux passes through the 
tube at right angles to the axis of the beam, the latter will be deflected 
at right angles to the direction of the flux, i.e. at right angles to a line 




2 3 4 5 6 78310* 2 3 4 5 8789/0 3 2 3 4 5 6789/0* 

Frequency, in cycles per second 
Fig. 190. Curve demonstrating the Hollmann effect 

passing through the centres of the two coils. The amplitude of the 
deflection may be calculated from the equation: 

_ - . 0-4LIIN 
L>enection= —-_ approximately 

(assuming there is no flux leakage—in practice, it is usual to employ 
a constant of about 0-3 instead of 0-4) 

where L is the distance from the centre of the pair of coils to the screen, 
/ the length of that part of the beam which is in the magnetic 

field, 
J the current in the coils, 
N the total number of turns through which the current / flows, 

and 
V the final anode potential. 






APPENDIX I 



273 



All the distances are to be measured in centimetres and the current 
in amperes. In this case the deflection is proportional to the flux, or 
for air core coils, to the current in the coils and inversely proportional 
to the square root of the final anode potential. 

The Hollmann effect also occurs with magnetic deflection, but is 
unlikely to be met with in practice, since the amount of power re- 
quired to deflect the spot electro magnetically at such a high frequency 
is generally too great to permit the method to be employed. 

BRIGHTNESS MODULATION 
Brightness Modulation of the Trace 

Brightness modulation, apart from the blacking-out of a fly-back, 
may be used in two ways : 

(a) As in television, where the whole of the necessary information is 
obtained by examination of the comparative brightness and 
position of different parts of the image. 

(b) As a means of decreasing the variations of brightness along a 
trace which contains large variations of writing speed such as, 
for example, a trace of a very sharp pulse or of a saw-tooth with 
a very rapid fly-back. 

Brightness modulation of the trace may be obtained by applying a 
potential having the desired characteristic either to the grid or cathode 
of the cathode ray tube. This modulation of the brightness of various 
parts of the image forms the basis of the intensity modulation method 
of television reception with a cathode ray tube. 

Brightness modulation is of particular value when it is desired to 
examine an impulse having a steep wave-front or where the signal is 
in the form of a pulse which lasts for a very short time and recurs at 
relatively long intervals. When the wave-front is steep, the velocity of 
the deflected trace at this point is high and the image will be very dim 
compared with the remainder of the trace. If only the impulse is 
allowed to appear, the remainder of the trace being blacked out, the 
brightness of the required portion can be greatly increased with a 
corresponding increase in the visibility. By employing a differentiating 
circuit (see Appendix IV), the brightness may be increased for the high 
frequencies (steeper wave-fronts) and by proper adjustment of the 
amplitude of the signal from the differentiating circuit, together with 
the value of the time constant, the degree of change of brightness for 
the different parts of the trace may be controlled. 

It should be noted that, unless special arrangements are made, only 
signals of one polarity can be brightened. 



274 



TIME BASES 



Unwanted Brightness Modulation 

In those cases where the earth connection is made to a point at, or 
near, the positive rail of the H.T. supply to the cathode ray tube, 
brightness changes occur due to the unavoidable fluctuations in the 
potential of the supply mains. For example, a i per cent change in the 
supply mains will result in a 50-volt variation in a 5, 000- volt H.T. 
supply system. The potential across the condenser which is used to 
feed the modulator grid or cathode of the cathode ray tube must change 
by 50 volts, and this results in a change in the current passing through 
the grid or cathode resistance and, hence, there arises a change in the 



HT+o- 



n I 
f i . T 

r\ j-t T ' 




o Receiver 
H.T.+ 




.q From 
r\ Receiver 



Modulator I"' 



Jenkins' method of avoiding unwanted brightness modulation 



trace brightness. If, however, the cathode and the grid can both be 
arranged to vary in potential by the same amount and in the same sense, 
no effect on the brightness of the trace will result. Jenkins* has 
developed a circuit, shown in Fig. 191, to produce this effect. It will be 
noticed that, with Jenkins' circuit, any change in the potential appearing 
across the cathode ray tube potential divider is applied equally to both 
electrodes through symmetrically arranged circuits. In particular, the 
symmetrical arrangement of the bias control circuit should be noted. 
The resistances R 2 must be equal and should be small in value 
compared with i? 3 . In general, the impedance i? 4 in the anode circuit 
of the modulating valve is small compared with the grid or cathode 
resistances R lt and, hence, it is normally unnecessary to duplicate it 
in the feed line to the other condenser C 2 . 
The following values may be employed for the component parts : 

* J. W. Jenkins: "The Development of C.H. Type Receivers for Fixed and Mobile 
Working," Journ. LE,E„ 1946, 93, Part IIIA, p. 1123. 






APPENDIX I 



275 



R x i-o megohm each 

R 2 0-25 megohm each 

i? 3 i-o megohm 

#4 small compared with R x (say 0-1 megohm maximum) 

C t o-i uF. 

C 2 dependent upon the lowest frequency to be transmitted. 
The two condensers must be of equal value. 

BEAM VELOCITY 
It is occasionally required to know the velocity of the beam which is 



^=3x10 



10 



V 1 V2Xio- 6 F+i/ 



centimetres per second, 



,2Xie>- 6 V-\-: 
where V is the potential of the final anode. 

For values of V not much greater than 10,000 volts a more simple 
equation may be used. This is 

a =5-05 xio'VK centimetres per second approximately. 
The difference between these two equations is that the latter does not 
take into account the change of mass of the electron with velocity. 

DISTORTION 

There are two generic forms of distortion in the cathode ray tube and 
these are due to the following causes : 

(a) Bad design or construction. 

(b) Incorrect use. 

Of the faults due to bad design or construction of the cathode ray 
tube something has already been said on page 261. 

Another effect, which does not occur with modern tubes, was dis- 
covered by Puckle in 1929, but he was unable to explain the cause. 
When a steady potential is applied to a deflector of a cathode ray tube 
(not the modern variety), the spot moves across the screen and slowly 
returns to a position which is nearer to the centre of the screen. It was 
not until 1934 that Wheatcroft* was able to explain the cause as being 
due to a polarizing layer which forms upon the surface of the deflector 
plate, resulting in a reduction in the potential gradient as the layer builds 
up and, therefore, reducing the deflection of the spot. The polarizing 
layer disappears if the potential is removed from the deflector plate. 
At the time, it appeared that cathode ray tubes could not be used to 
measure potentials which were steady or which change at a very slow 
rate. It was, however, discovered by Bedfordf that if the deflector 

»E, Wheatcroft: Sheffield University, Sheffield. 

fL. H. Bedford: Marconi's Wireless Telegraph Company, Ltd., late A. C. Cossor, 

Ltd. 



I 



276 



TIME BASES 



plates were made of molybdenum, the effect does not occur. The effect 
is known as the Wheatcroft effect. 

Those faults which are due to incorrect use of the tube are caused by 

(a) The use of incorrect potentials. 

(b) The use of asymmetric potentials. 

The Use of Incorrect Potentials 

Bad focusing of the spot can, obviously, result from incorrect adjust- 
ment of the focus potential. It can also result if the instantaneous value 
of the grid bias potential becomes too low. 

Asymmetric Distortion 

It is important to note that two forms of distortion occur in the 
cathode ray tube when the deflector plate potentials are applied in an 
asymmetric manner. These distortions are Deflection Defocusing and 
Trapezium Distortion. 

Deflection Defocusing 

The application of a deflection potential to one plate of a pair brings 
about a progressive defocusing of the spot as it travels to one side of 
the screen. This can largely be avoided by employing symmetrically 
applied deflecting potentials. 

Trapezium Distortion 

If the pair of deflector plates farther from the lens system (X plates) 
are asymmetrically supplied, a cylindrical lens is produced between the 
two pairs of deflector plates and between the X plates and the screen. 
If the beam is initially deflected by the Y plates, then, when the live 
X plate is positive, the beam is deflected inwards in the plane of the 
X plates by the cylindrical lens both before reaching and after passing 
the deflector plates. The field round the X plate, in the case of asym- 
metric deflection, or round both X plates if they are operated in 
push-pull, but with their mean potential different from that of the final 
anode, is shown in Fig. 192. Assume the X plate to be positive with 
respect to the anode, then, for a beam deflected at the point A, there 
will be an acceleration at B in the direction of the arrow. Since this 
force contains a component towards the axis of the tube, the beam 
deflection is reduced. Upon emerging from the plates, a deceleration 
occurs at the point C, and in this case also there is a component towards 
the axis so that a further reduction in deflection is suffered by the 
beam. The arrows denoting the directions of the accelerating and 
decelerating forces are drawn normal to the equipotential lines. 









APPENDIX I 



277 



The X plate becomes negative at one end of the time base trace 
and positive at the other end, with the result that the deflection 
due to the Y plates is decreased when the X plate is positive and 
increased when the X plate is negative. For this reason, the area 
enclosed by the curve portrayed on the screen is trapezoidal as shown 
in Fig. 193. A tube structure which completely overcomes this effect 



4- 




X 



Finaf 
Anode 



Fig. 193. Diagram showing how 
trapezium distortion is produced 



Finaf 
Anode 



Fig. 194. Diagram showing the 
shape of the anti-trapezium dis- 
tortion screen T and of the X- 
defiector plates 




'X- o 
Fig. 193. Diagram illustrating trapezium distortion in a cathode ray tube 

has been designed by Fleming- Williams,* and is shown diagrammati- 
cally in Fig. 194. 

In this arrangement, the leading edges of the X plates are curved 
and a special shield T, also curved, is interposed between the X and Y 
deflecting systems. This shield, which is connected to the final anode, 
prevents the shape of the field round the X plate from becoming 
changed as the potential on the Y plate is varied, since it acts as a uni- 
potential screen. The reduction of the deflection, when the beam 
enters and leaves the X plates, still occurs but, since it is now 
of fixed magnitude, the trapezium distortion effect disappears. This 

*B. C. Fleming-Williams: "Trapezium Distortion in Cathode Ray Tubes," Wireless 
Engineer, 1940, 17, p. 61. 



278 



TIME BASES 



arrangement functions correctly with either X plate earthed and, 
if push-pull X deflection is employed, the mean potential of the X plates 
may be positive or negative with respect to the final anode. An internal 
view of an anti-trapezium distortion tube is shown in Fig. 187. In the 
case of Fig. 182, however, which is also an anti-trapezium distortion 
tube, the shield T has been partially cut away to enable the splitter 
plate to be more clearly seen (see note T on page 268). 

Astigmatism 

There is a further distortion, known as astigmatism which, however, 
is not due to asymmetry of the deflection potentials. The term "astig- 
matism" has the same meaning in electron-optics as in light-optics, 

n Horizontal ~ 
X Def/ectiofl i 
■ ■ Potential ■ ■ 



Vertical 

Deflection 
Potential 




-O- 



.IP 

Fig. 195. Anti-astigmatism circuit 

Le. it denotes the condition where good focus is obtained along one 
axis together with lack of focus along an axis at right-angles to the 
first. The effect may be introduced by misalignment of the various 
parts of the electron lens system or by misalignment of the deflector 
plate system with respect to the lens. Furthermore, the presence of 
certain electrostatic fields, which occur when the mean potential of the 
deflector plate system with respect to the final anode is incorrect, may 
introduce the effect. Deflection defocusing also introduces an astig- 
matic effect. 

The amount of the error in alignment which introduces astigmatism 
is so small that correction is necessary with some tubes and a method, 
due to Puckle,* of overcoming the effect will therefore be described. 

*0. S. Puckle: "Electrostatic Deflection," Wireless World, April 28th 1018 
*2, p. 375. K yj * 






APPENDIX I 



279 






In order to compensate for astigmatism it is necessary to arrange 
that the mean potential of each pair of deflector plates bears some 
particular relation to the potential of the final anode and this often 
requires that the mean potential of each pair of plates must be separately 
adjusted. In this way, it becomes possible to introduce two variable 
cylindrical lenses into the system, and these lenses may be adjusted to 
neutralize those which are the result of misalignment or incorrect mean 
potential. Since the effect of astigmatism increases with the amplitude 
of the deflection, it becomes particularly important to compensate the 
effect when the deflection is large. A circuit which may be employed 
for this purpose is shown in Fig. 195. It is necessary to ensure that the 
anti-astigmatizing potentials are free from ripple, either across the 
sources or between each source and earth, as, otherwise, the correcting 
circuits will introduce disturbances of the position or focus of the spot. 

The Measurement of the Distortion in a Cathode Ray Tube 

Many workers have devised circuits for measuring the distortion 
inherent in a cathode ray tube and for making measurements to obtain 
a figure of merit for the quality of the tube. These systems usually 
provide a modulated raster in which extreme care is taken to obtain 
linearity of scanning in each direction. The modulator of any such 
system can be used to provide overall checks for a complete television 
receiver or oscillograph equipment. 

A relatively simple equipment for carrying out these measurements 
has been described by Spooner.* The type of pattern obtained on a 
cathode ray tube is shown in Fig. 196, while Fig. 197 shows the method 
of assessing the distortion by means of a series of fine lines engraved 
at half-centimetre intervals on a celluloid sheet. By placing this sheet 
in front of the cathode ray tube, the degree of linearity in the horizontal 
direction may be measured and, by rotating the sheet through 90 , 
vertical linearity may also be determined. Barrel or pin-cushion distor- 
tion may also be assessed by means of the sheet, while the existence, or 
otherwise, of deflection defocusing may be made evident by measuring 
the diameter of the spot at various parts of the cathode ray tube screen 
with the aid of a microscope. This should be carried out with different 
values of beam current. 

The modulated raster shown in Fig. 196 is produced by the circuit 
shown in Fig. 198, which operates in the manner described below. 

The valves V 1 and V 2 form an asymmetric multivibrator whose 
frequency may be varied between about ten and fifty times the line 
scanning repetition frequency. The number of spots in the horizontal 

* A. M. Spooner: "C.R. Tube Quality Measuring Apparatus," Journ. Television 
Society, 1946, 4, p. 251. 



280 



TIME BASES 



direction is thus adjustable so that they may be made to approximate 
to a half-centimetre apart, for convenience in measurement. This 
frequency variation is brought about by adjustment of R 5 , R Q and C 2 . 
The multivibrator is synchronized by positive line-frequency pulses 
taken from the raster time base generator. This generator must have 
particular attention paid to the linearity of the two time bases. 

The output of the multivibrator is amplified by the valves V z and F 4 , 
the time constant of /? 7 C 4 being short so that the pulse producing the 
dots has a duration of less than one picture element. 

The vertical series of spots is produced by raising the grid potential 
to a value which is insufficient to cause the spot to appear unless 
the short pulse from F 4 is superimposed. To obtain this effect, it 
is necessary to have a positive square wave of duration equal to, or 
just greater than, one line period, followed by a negative signal period 
lasting until the next horizontal line of pulses in the same scan is due 
to appear. This square wave is generated by the valves F 9 and V 1Q . 
In this arrangement, V x is a triode-pentode which functions as a multi- 
vibrator. R B2 is used to adjust the repetition frequency to one-third 
of line frequency by means of an oscillograph connected to the test 
point near the condenser C 22 . The circuit is synchronized by the 
positive synchronizing pulses from the raster generator. V % is also a 
multivibrator, synchronized from F 10 and adjusted by means of R z9 
and C 17 so that it runs at one-fifteenth of line frequency, i.e. at one-fifth 
of the frequency of V 1Q and so that the negative pulse lasts for just over 
one line length. The diode V s has its anode potential adjusted by means 
of the potentiometer 2? 25 to provide a flat top to the negative part of 
the output wave. This negative portion of the square wave, the ampli- 
fied short-duration pulse from F 4 and the line and frame positive 
blanking signals from the raster generator, are fed to F 7 , F 5 and F 6 , 
respectively, into a common anode load R l8f where they are mixed, and 
fed via C n to the cathode ray tube. 

When the switch S t is alone closed, twenty-seven horizontal lines 
appear. When S z is alone closed, a number of vertical lines appear, 
the number depending upon the adjustment of R 5 , R G and C 2 . When 
both 5 2 and S 2 are closed, a series of dots appears on the raster. Closure 
of the switch S 3 injects the line and frame blanking signals and thus 
provides fly-back suppression. A further switch S 4 reduces the value 
of the anode load of the valves F 5 , F 6 and V 7 . 

Non-linearity of the line time base appears as non-equal spacing of 
the vertical lines as shown at i in Fig. 197, while non-linearity of the 
frame time base gives non-equal spacing of the horizontal lines as shown 
at 2 in the same figure. The celluloid screen may be rotated through 
90 for measuring the degree of non-linearity of the frame time base. 






APPENDIX I 



281 




Fig. 196. Modulated Raster produced by the circuit of Fig. 198 



Tube 



Celluloid Sheet 



between 
fines 




Picture 
limits 



Fig. 197. Gauge for assessing deflection distortion in a cathode ray tube 



282 



TIME BASES 



APPENDIX T 



283 




Barrel, or pin- cushion, distortion may be checked by noting the 
deviation from a straight line of the outside lines of dots when compared 
with the celluloid screen markings as shown at 3 in Fig. 197. 

The following component values may be employed for the circuit 
of Fig, 198. 



S 

I 

.O 

B 

.'- 

u 



R 3t R4 

R$> -^9 
#10 
*11 

^M2> ™M 
^14 

^10 

R 
R 



IN 

19 
R20 
R 2 l 

Una 



^23 
^24 
^25 
^26 
^27 
^28 
^29 
^30 
^31 



H 



:;:! 



R. 



M 



4,800 ohms each 
12,000 ohms each 
50,000 ohms 
5,000 ohms 
o*47 megohm 
12,000 ohms each 
1,200 ohms 

1-0 megohm 
25,000 ohms each 
15,000 ohms 

100 ohms 

150 ohms 
2,200 ohms 
6,000 ohms 

o-i megohm 

150 ohms 

100 ohms 

o-i megohm 

100 ohms 

150 ohms 

0-5 megohm 
33,000 ohms 

1-2 megohms 
33,000 ohms 

0-7 megohm 

o*i megohm 
33,000 ohms 
50,000 ohms 
33,000 ohms 
i-2 megohms 



Ci 

c 2 
c 3 
c 4 



^1Q> ^-11 
^-12 
^-13> *--14 

C 17 

£-19 
^-20 
^21 
L-22 



47O fl[xF. 
5O jJ-JiF. 
56 (ZfiF. 

6-8 wj-F. 

o-i tiF. 

8-2 |A|xF. 

100 jajjiF. 

o-i (iF. 

100 [ifiF. 

0-47 [aF, each 
6,800 {ifiF. 

47 (xF. each 
6,800 fi.(iF. 
1,000 p-tiF. 

100 wlF. 

8-2 (xfiF, 

100 jxjJtF. 
6,800 y.\>.F. 

180 ixfiF. 
10 tt(iF. 



Fi-F, EF50 

K 8 EB34 

V 9 , F 10 ECH35 

F n AZ31. 






Deflection defocusing is checked by observing whether the dots at 
the outside edges of the raster are circular and of the same diameter 
as those at the centre. A microscope may be needed for this latter 
purpose. The change of spot size with varying values of beam current 
may also be observed with the microscope, but the use of a separate 
circuit is of value if quantitative results are required. The appropriate 



284 



TIME BASES 






circuit is shown in Fig. 199. In this circuit, a microammeter is used to 
indicate the value of the beam current and a valve voltmeter indicates 
the value of the positive bias applied to the cathode of the cathode 
ray tube. 

When the input terminal is earthed, a raster of uniform brightness 
appears due to the application of the output from the raster generator 
to the deflector coils or plates of the cathode ray tube. Variation of the 
cathode bias potentiometer R 4 changes the beam current and the bias 
potential may be read off the appropriate meter. 



Input 



5000 V. 







Brightness 
Control on 
Receiver 



Fig, 199. Circuit for determining the change of spot size with heam current 

When the grid is unearthed, the output from the quality measuring 
circuit (modulator) of Fig. 198 provides a series of dots on the raster, 
and the peak potential of the pulses which produce the dots may be 
measured by means of the diode circuit of Fig. 199. Adjustment of the 
potentiometer R 2 controls the value of a backing-off potential, and, if 
this is altered till current just ceases to flow in the microammeter 
connected across C 2 , the potential indicated on the voltmeter V b is the 
peak potential between the cathode and grid of the cathode ray tube. 

Corresponding readings of the peak input potential, beam current 
and spot diameter may be made. It is more satisfactory to make spot 
diameter measurements with a modulated raster since the fluorescent 
screen is not overloaded, as is likely to be the case when only a single 
spot is employed. 

The following values may be employed for the components of 
Fig. 199: 



APPENDIX I 



285 



#5 J 

c 2 



i-o megohm 
3,000 ohms 

as required for the particular cathode ray tube 



0-64 nnF. 
o- 1 nnF. 

T4D. 



THE APPLICATION OF SHIFT POTENTIALS 
There are many occasions when it becomes advantageous, or essential, 
to employ shift potentials in order to provide a means of shifting the 
initial steady position of the spot away from the centre of the screen. 
Occasionally, it is desirable to employ a trace which is considerably 
longer than the diameter of the screen and to shift the trace sideways, 
or up and down, in order to examine any particular part of the image 
at will. The use of an extended trace permits a microscopic examina- 
tion of the detail in the image. To obtain the maximum benefit from 
the operation, the deflecting potentials should be more than sufficient 
to provide full screen deflection in each direction. 

When the deflecting potentials are applied to the plates through 
condensers with leaks taken to the final anode, the mean potential of 
the applied wave-form will correspond with zero deflection. 

If the mean potential of a wave-form does not coincide with the lme 
which is half-way between the positive and negative peaks, the use of 
shift potentials will permit the image to be centralized upon the screen. 
The mean potential of any wave-form is that potential at which the 
area under the curve is the same for the positive part of the wave as 
it is for the negative part. This may be appreciated by reference to 
Fig. 218 (differentiation), from which it will be noticed that the 
positive and negative potential excursions may be very unequal. 

In the case of an ideal saw-tooth wave-form, the mean potential lies 
half-way between the positive and negative peaks but, if there is a time 
interval between successive traces, the mean potential approaches the 
potential which exists during the interval. This tendency becomes 
more pronounced as the interval-trace time ratio increases. Therefore, 
when using a single-stroke time base or one in which the duration of 
the trace is a small fraction of the repetition time, it is usual to bias 
the spot initially to one side of the screen in order that the trace may 
be centrally located. 

Although it is possible to obtain a shift of the image by means of 



286 



TIME BASES 



a magnet or electro-magnet, it is usual with electrostatic-deflection 
tubes to employ electrostatic shift devices. A battery may be employed 
to provide the necessary shift. When push-pull deflection is used, two 
batteries are required and two further batteries will be necessary for 
the other direction of deflection. This method is cumbersome and adds 
considerably to the stray capacitance of the deflecting systems, but it is 
often the only practicable method. It is employed when it is desired 
to use the cathode ray tube as a peak voltmeter to indicate actual rather 
than difference potentials. 

If the deflection potentials are applied to the plates via condensers 
and resistances, it is convenient to apply the shift potentials at the 
earth ends of the resistances R x and R 2 , as shown in Fig. 200. 



AWWW 




WW 1 1 It 

I — vvwvwvw — 



Ri 



YShift\ 



jWWAW- 



:^4 



HO + 



F/oating 

Astigmatism 
q Correction 



-O- 



Flg, 200. Push-pull circuit for shifting the position of the trace and 
providing correction for astigmatism in one deflectional direction only 

The circuit in this figure shows a method of obtaining the astigmatism 
and shift potentials from one source but, if it is required to apply 
similar potentials to both the X and Y plates, two such circuits with 
separate steady supply potentials will be necessary since, otherwise, 
independent astigmatism correction will not be obtained. 

The following circuit values may be employed: 



R 3 R, 



H.T. 



i to 5 megohms each 

0-5 megohm each (ganged linear potentiometers) 

0-5 megohm 

say 1 u.F. each 

say 200 to 500 volts or more (depending upon the 

sensitivity of the cathode ray tube). 

The H.T. supply potential must be "floating" with respect to earth, 
and it must be well smoothed so that there is no appreciable ripple 
across the supply rails or between either rail and earth. The slider of 
the potentiometer R & will normally lie near the centre but, by moving 
it to either side, astigmatism correction may be applied. The potentio- 
meters i? 3 and /? 4 , which are ganged together to form the shift control, 






APPENDIX I 



287 



must both have the same grading and, for this reason, it is advisable 
to specify the linear law. However, a close tolerance on resistance 
value is not necessary. It will be noticed that one of the potentiometers 
is connected in the reverse sense with respect to the other. This is 
necessary so that when one deflecting plate is made more positive 
with respect to earth, the other becomes more negative. If the law of 
the two potentiometers is not the same, the shift potentials applied to 
the two plates will not always be equal and, if this lack of equality is 
more than a certain amount, defocusing of the spot will take place. 

If the deflecting potentials are not applied in push-pull, the shift 
potential should be applied only to the same plate as the deflecting 
potential, the unused plate being connected directly to the final anode 
of the cathode ray tube. 

It will be found that the rotation of the shift control is followed by 
a slow movement of the image on the screen ; this is due to the necessity 
for changing the potential across the condensers C^ and C 2 and, if it is 
required to reduce the time delay, it will be necessary to consider 
reducing the time constant, but due regard must be paid to the possi- 
bility of producing other undesirable effects. For low-frequency work, 
where it is not permissible to employ short-time constants, this difficulty 
may be overcome by dispensing with condenser couplings and incor- 
porating the shift device in the time base itself (see page 222). In the 
case of the signal deflection, one may avoid condenser couplings by 
using a direct coupled push-pull amplifier, in which case the grid bias 
may be employed to control the degree of shift. 

The necessity for the use of two separate steady potentials may be 
eliminated by the use of the circuit of Fig. 201. This circuit provides 
independent shift and astigmatism correcting circuits, but the supply 
potential must be double that required for Fig. 200 when used with 
the same cathode ray tube operating with the same potentials. 

In this circuit, it will be noticed that the astigmatism correction 
potentiometers are so arranged that the total resistance in circuit 
remains constant provided that the potentiometers are complementary. 
This means that, as the controls are rotated, resistance is removed from 
one arm of the feed circuit and an equal resistance introduced into 
the other. This requires a close tolerance on the resistance value of 
the potentiometers. In this way the mean potential of one pair of 
plates is raised or lowered with respect to the final anode which is 
connected to the junction of the two resistances i? 18 and R 14 . As 
before, the shift potentiometers are connected in opposition so that as 
one plate is made more positive the other is made more negative. 

If, in either circuit, the potentiometers which are ganged together 
have dissimilar laws, variation of the shift controls will disturb the 



288 



TIME BASES 



astigmatism adjustment and, in the case of Fig. 201 only, adjustment of 
the astigmatism control will disturb the position of the image. 

The following values may be employed for the circuit of Fig. 201 : 



R 1 R 2 R^Rq 

R^R,R 7 R 8 
RqRiqRiiRiz 

^13^14 

*•" IV 2T- 3^ 4 

H.T. 



1 to 5 megohms each 

1 megohm each (ganged linear law) 

0-5 megohm each (ganged linear law and small 

difference in resistance) 

o-i megohm each 

say 1 fiF. each (dependent upon frequency) 

4 to 8 jtF. each 

say 400 to 1,000 volts (dependent upon the cathode 

ray tube sensitivity). 



Y Astigmatism 
Correction 




-0 + 



*J3l 



R 



H: 



I 



" s Floating 
D.C. 

r- Supply 



-o— 



X Astigmatism 
Correction 



Fie. 201. Push-pull combined shift and anti-astigmatism circuits for use 
in both deflectional directions 

In the case of Fig. 201, the supply potential is, as shown, earthed at 
the junction of the two resistances R 1S and i? 14 , and this point is also 
connected to the final anode of the cathode ray tube. Many variations 
of the above circuits, which are due to Puckle, are possible. 

THE USE OF ALTERNATING POTENTIALS FOR THE 

OPERATION OF A CATHODE RAY TUBE 

Bedford* has employed alternating potentials for the operation of a 

*L. H. Bedford: sometime A. C. Cossor, Ltd., London, now Marconi's Wireless 
Telegraph Company Ltd., Chelmsford. 



APPENDIX I 



289 



cathode ray tube in connection with certain apparatus used in con- 
junction with a radar circuit. In this apparatus, the potentials applied to 
the deflecting plates vary in phase and amplitude, but they consist only 
of 50-cycle sinusoids. The varying phases and amplitudes require to 
be indicated by the position of a spot on the surface of the fluorescent 
screen. The circuit adopted is shown in Fig. 202. It will be seen that 
all the supplies to the cathode ray tube are alternating potentials 
obtained directly from a transformer operating from the 50-cycle mains. 
The potential across the entire secondary winding (AD) of the trans- 
former is approximately 1,100 volts R.M.S. The potential across the 




Fig. 202. Bedford's method of using alternating potentials for 
operating a cathode ray tube 

winding BD is a focus potential and is adjustable by means of the 
potentiometer R 2 . ^ ne network R A R iy which is connected across the 
winding CD, is fed with a potential of approximately 100 volts R.M.S. 
and is applied to the modulator grid via the condenser C 1 and resistance 
R v The diode V Xt in parallel with the resistance R lt causes the con- 
denser C x to charge until current flows in the diode only during the 
extreme positive peaks of the wave-form. The spot is thus only able 
to appear for a very short period of the mains-frequency cycle and 
therefore a spot of small dimensions is produced. A brightness control 
is provided by the adjustment of the potentiometer i? 4 . 

THE USE OF THE CATHODE RAY TUBE 
A few details to which attention must be paid are given below: 
Since the cathode ray beam can be deflected by a magnetic field it 
is important to ensure that the deflections observed are not affected by 
the presence of unsuspected stray fields, for, if this occurs, the infor- 
mation extracted from the image may prove to be erroneous. 



290 



TIME BASES 



Great care must therefore be taken to avoid the presence of un- 
wanted magnetic or electric fields in the neighbourhood of the tube. 
The presence of mains-frequency fields can generally be detected by 
short-circuiting the deflector plates to the final anode of the tube and 
by rapidly moving a permanent magnet back and forth in the neighbour- 
hood of the deflector plates. This will deflect the beam in one direction 
and any mains ripple which may be present will become visible. 

The presence of a steady magnetic field may be detected by gradually 
reducing the H.T. potential with the deflector plates short-circuited. 
The reduction of potential increases the deflectional sensitivity and the 
spot will therefore move farther from the centre of the tube. By moving 
the tube in space until maximum deflection is obtained, the direction 
of the field can be determined. The beam may be considered as an 
electric current flowing from screen to cathode without a conductor 
and, like all currents, it sets up a magnetic field around it. This field 
takes the form of a cylinder which is concentric with the beam and the 
direction of the flux will be clockwise as viewed from the screen end 
of the tube. Hence, an external magnetic flux passing from left to right 
across the tube will deflect the beam downwards. 

Having found the direction of the external field, the source can often 
be detected and removed from the area of the cathode ray tube. If the 
external field is that of the earth and the tube is affected thereby, the 
normal solution is to cause the flux to by-pass the tube with the aid of 
a magnetic screen. Alternatively, the tube may be so oriented that the 
external field lies along the length of the tube, but it must be remem- 
bered that if the tube be so oriented, the beam will be unaffected by 
this field only so long as it remains undeflected. The deflected beam 
will execute a slow spiral round the axis of the field. For a three- 
centimetre deflection with 1,000 volts on the anode, the beam will rotate 
through an angle of approximately three degrees in England. Another 
method is to arrange two external bar magnets which may be placed, one 
on each side of the tube, so as to cancel out the undesired field. These 
bar magnets should be powerful ones, placed at a distance, rather than 
weak ones near the tube since, in this way, the lines of force will be 
more nearly straight and will therefore compensate the unwanted field 
over a larger portion of the length of the tube. The correct positions 
for the bar magnets should be found by trial and error with the 
deflector plates short-circuited to the final anode. 

The presence of extraneous electric or magnetic fields in the lens 
system of the tube will generally bring about defocusing and may even 
deflect the beam sufficiently to prevent it from reaching the screen. 
Care should therefore be taken to see that transformers and chokes are 
well removed from the tube especially from the cathode end since, at 



APPENDIX I 



291 



this end, the electron velocity is lower and the beam is therefore more 
easily deflected. 

Stray fields from transformers and chokes may be reduced by 
designing them on a low flux density basis and by orienting them with 
respect to one another and to the tube so that the stray fluxes oppose 
one another and lie, so far as is practicable, along the length of the tube. 
In this way, the resultant flux in the neighbourhood of the tube will 
be reduced to a minimum and the effect of this minimum flux is further 
reduced by correct orientation. This adjustment is best made by trial 
and error and, of course, it can only be carried out with transformers 
and chokes operating on the same supply frequency. 

Another method of reducing the effect of stray magnetic fields is, as 
already stated, to surround the tube with a magnetic screen. Mu-metal,* 
permalloy, etc., may be used for this purpose. The permeability of 
mu-metal is extremely high, especially for relatively low values of the 
magnetic flux. The screen, if connected to earth, will also greatly reduce 
the effect of electric fields. If a single mu-metal screen proves insuffi- 
cient to remove the effect of the interfering field and the source of the 
field cannot be removed or otherwise attenuated, a second mu-metal 
screen can be arranged to surround the first with a gap of about £ inch 
between them. As a result, the flux must pass through two air-gaps to 
reach the cathode ray tube and is thus considerably attenuated. 



Cross -talk 

If electromagnetic deflection is used for both the X and Y axes of 
the tube, it is obviously important to ensure that no coupling exists 
between the pairs of coils which might permit the current in one pair 
to induce a current in the other. If this were to happen, the current in, 
say, the X coils might produce a spurious deflection in the Y direction, 
or vice versa, and so produce an inaccurate trace. 

Similarly, when electrostatic deflection is used, excessive stray 
capacitance between the plates and/or the leads connected thereto may 
cause mutual interference of the traces. The following are the con- 
ditions which make electrostatic cross-talk possible: 

Exciting Circuit 

(1) High deflection potential. 

(2) High deflection speed. 

(3) Unbalanced stray capacitance between the exciting circuit and 
each of the plates of the excited pair. 

* W. F. Randall: "Nickel-iron Alloys of High Permeability with Special Reference 
to Mu-metal," Journ. I.E.E., 1937, 80, p. 647; G. A. V. Sowter: "The Properties of 
Nickel-iron Alloys and their Application to Electronic Devices," Journ. Brit. I.R.F.,, 
1041. 2, p. 100." 



2Q2 



TIME BASES 



Excited Plate Circuit 

(a) High impedance across the plates. 

(b) Unbalanced stray capacitance between each plate of the excited 
pair and the exciting circuit. 

Of these conditions, 2 and 3 are closely related while 3 and b are 
identical but expressed in terms of the two circuits. The question of 
the impedance of the excited circuit is of fundamental importance since 
the phenomenon depends primarily upon this quantity. When the 



■■fl— www-* ► 



i?! —R z orZ l =Z 2 
R 2 = R^orZ s =Z t 







fxcitmg 
Potential 



Fig. 203. Circuit for the compensation of cross-talk in a cathode ray tube 

impedance across the excited circuit is negligible compared with that 
of the stray capacitance between the pairs of plates at the maximum 
frequency concerned, the effect does not appear. Moreover, if the 
stray capacitance from the exciting pair to each of the plates of the 
excited pair is equal, the effect is again non-existent because, in this 
case, a balanced bridge circuit is formed. All bridge circuits, in which 
the products of the impedances of opposite pairs of arms are equal, are 
said to be balanced, and have the unique property that a potential 
applied between opposite corners produces no difference of potential 
at the other pair of opposite corners. Thus, in the circumstances 
depicted in Fig. 203, any potential applied to the X plates has no effect 
on the Y plates, and vice versa. 

It will be seen that in the figure the Y plates are the excited ones and 
the condensers C 3 and C 4 are adjustable condensers arranged to com- 
pensate the stray capacitances C t and C 2 . If C^ is equal to C 3 and 



APPENDIX I 



293 



C 2 is equal to C 4 , either pair of plates may be energized without 
mutual interaction. 

The best method of preventing cross-talk is as follows: Apply the 
appropriate signal to one deflector plate, say, an X plate, the other 
X plate being connected to the final anode. The signal to the Y pair 
should be removed by disconnecting the deflector plates and con- 
necting them to the final anode through separate resistances of about 
2 megohms each. A small variable condenser of a maximum value of 
about 10 micromicrofarads should be connected between the exciting 
X plate and one of the Y plates; the condenser should then be 
adjusted until the deflectional component in a direction at right angles 
to the normal is removed. The correct excited Y plate to which the 
condenser should be connected is best found by trial. 

When this has been done, the same procedure is carried out for the 
other exciting plate, using a separate condenser. When the condensers 
have been correctly adjusted they should be left in position. The 
oscillograph may then be used without the fear of difficulty due to 
cross-talk, providing the connecting leads are reasonably short and 
well spaced. If this is not the case the above-mentioned tests should 
be carried out with the leads remaining connected to the deflector 
plates. 

It is assumed that at least one pair of plates will be operated in push- 
pull, otherwise the only solution is to employ low impedance deflector 
plate circuits. The method is due to Puckle. 



The Use of a Cathode -follower Probe 

When a cathode ray tube is employed to examine the wave-form at 
various parts of a circuit, the additional capacitance of the deflector 
plate system and the connecting leads thereto is sometimes sufficient 
to disturb the normal operation of the circuit. In these circumstances, 
it is advisable to take advantage of the property of a cathode-follower 
which results in a greatly reduced input capacitance. The circuit and the 
method of using it are discussed on pages 129 et seq. If the value of the 
cathode bias resistance is made sufficiently high so that the cathode 
follows the grid to within, say, 1 per cent, the input grid to earth 
capacitance is reduced to approximately 1 per cent of the grid to 
cathode capacitance. Since the impedance looking into the cathode load 
of a valve operating as a cathode follower is approximately 150 to 200 
ohms depending on the value of the bias potential, but otherwise 
independent of the value of the cathode load resistance, the effect of 
the shunt capacitance, due to the deflector plates, connected across the 
cathode load resistance is negligible. The grid to anode and grid to 
screen capacitances, however, remain as before. 



294 



TIME BASES 
FLUORESCENT SCREENS 



Secondary Emission 

The beam current for the most part returns to the final anode of the 
cathode ray tube by secondary emission from the screen.* This 
secondary emission is the result of the bombardment of the screen 
by the electrons forming the beam, the screen taking up a potential 
usually of the order of 100 volts negative with respect to the final 
anode. The potential of the screen depends upon the fluorescent 
material of which the screen is made, the binder which is used for 
cementing it to the glass and the actual potential applied to the final 
anode. If the returning electrons are allowed to reach a deflector 
plate, the addition of a non-linear resistance load between the plate 
concerned and the final anode of the cathode ray tube results. This 
effect is, of course, more important when the deflector plate leak 
resistances are of high value. The non-linear resistance load may 
alter the deflectional sensitivity of the plates as the deflector potential 
changes sign, with consequent distortion of the trace. 

The reduction of the amount of secondary emission reaching the 
deflector plates is largely a matter of successful design of the tube, but 
it is also possible to aggravate the effect if the mean potential of the 
deflector plate system is made positive with respect to the final anode. 
The tube is normally coated, on the inside surface, with a layer of 
carbon particles which extend from the final anode to a point near the 
screen. This coating is connected to the final anode and serves to 
collect the secondary electrons which leave the screen and so prevents 
them from reaching the deflector plate system. A further method of 
overcoming the difficulty by coating the fluorescent material with a thin 
film of aluminium is described on page 295. 

Various fluorescent screen materials have been developed and the 
choice of a suitable screen is dependent upon the purpose for which the 
tube is to be employed. The main characteristics which may be 
selected by a choice of material are as follows : 

(a) Colour 

(b) Afterglow duration 

(c) Luminescent efficiency 

(d) Actinic efficiency, and 

(e) Saturation potential. 

The saturation potential is the limiting value of the final anode 
potential, beyond which, further increase of anode potential fails to 

*H. Moss: "Secondary Emission, Some Effects on Deflector Plate Characteristics 
of the Cathode Ray Tube," Wireless Engineer, 1941, 18, p. 309, 



APPENDIX I 



295 



produce any further increase in luminous output. A recent development 
is that of applying a thin film of aluminium over the surface of the 
fluorescent screen. This has two effects. Because the aluminium is 
connected to the final anode, and since its resistance is relatively low 
by comparison with that of the screen material, it prevents the accumu- 
lation of a negative charge and so enables the potential at which the 
screen is operated to be increased. As a result of this, the screen 
brightness can be increased at the higher values of anode potential. 
Furthermore, the presence of the aluminium reflects forward the light 
which would otherwise enter the body of the tube and be reflected 
from the walls of the tube. The general background light level is thus 
reduced and the high lights increased in intensity so that the contrast 
is improved. For this latter function, the aluminium must operate as 
a mirror and its surface, next to the fluorescent crystals, must be flat. 
It is therefore necessary to level the crystal surfaces before applying 
the aluminium and this is usually carried out by the application of a 
translucent plastic coating. 

The following table shows the screen materials which are available 
in the cathode ray tubes manufactured by Cossor, E.M.I., Mazda* and 
Mullard: 



Table VII 

Types of Cathode Ray Tube Screen 

Cossor Tubes 



Screen 
type 


Colour 


Approximate 
afterglow 
duration 


Luminescent 
efficiency 


Actinic 
efficiency 


Application 


J 
D 

G 

K 


Light 
Blue 

Green 

Blue- 
Green 

White 


10 micro- 
seconds 

10 milli- 
seconds 

10 seconds 
Negligible 


Medium 

Very high 

High 

Very high 


Very high 
Low 
High 

Medium 


Photography 
Visual work 

Visual and photo- 
graphic examin- 
ation of slow 
speed traverses 

Brilliant tele- 
vision images 



* Mazda tubes are made by the Edison-Swan Electric Company. 



296 



TIME BASKS 
K.M.I. Tubes 



Screen 
type 


Colour 


Approximate 
afterglow 
duration 


Lumin- 
escent 
efficiency 


Actinic 
efficiency 


Application 


ZnS.Ag + 
ZnS.CdS.Ag 


White 


100 microsecs. 


High 


Medium 


Television 


ZnBeS+ 
ZnS.Ag 

Willemite 

Mg F2. Mn 


White 
Green 
Orange 


50 millisecs. 
50 millisecs. 
30 seconds 


High 

nigh 

High 


Medium 

High 
Very low 


Television 

projection 
Visual 

work 
Radar 


W 


Yellow 


2 seconds 


Medium 


Low 


Radar 


AB 


Yellow 


3 seconds 


Medium 


Low 


Radar 


Mf 

M (improved)"!" 


Blue+ 
Yellow 
Blue-f 
Yellow 


30 seconds 

45 seconds 


Medium 
Medium 


Low 
Low 


Radar 
Radar 



Tubes marked thus f have double layer screens. They produce a blue 
flash followed by a yellow afterglow. 

It should be noted that since, with most screen materials, the afterglow is 
not exponential, the afterglow duration depends upon the duration and 
intensity of the excitation. The figures given above are therefore very 
approximate. 

Mazda Tubes 



Screen 
type 


Colour 


Approximntt- 
afterjxlow 
duration 


Luminescent 
efficiency 


Actinic 
efficiency 


Application 


P.I 


Green 


1/25 sec. 


High 


Medium 


Visual work 


P.2 


Greenish 
Blue 


45 sees. 


Low 


Low 


Visual observation 
of transients and 
slow repeating 
phenomena 


P-3 


Intense 
Blue 


Less than 
1/100 sec. 


Medium 


High 


Photography 


P-4 


White 


Very 
slight 


High 


Low 


Television 



APPENDIX I 

Mullard Tubes 



297 



Screen 
type 


Colour 


Approximate 
afterglow 
duration 


■ ■■ - — ■ — ! — 

Luminescent 

efficiency 


Actinic 
efficiency 


Application 


— 


Green 
Blue- 
White 


Medium 
Short 


High 

High 


Medium 
Medium 


Visual 
Television 



The following information is taken from papers by Hopkinson:* 

Table VIII 
Some Phosphors used for Cathode Ray Tube Screens 

With the exception of (Willemite), the names in brackets refer to intentionally 
added impurities, known as "activators" 



Phosphor type 


Colour 


Approximate 

relative 

luminous 

efficiency 

under 

cathode ray 

excitation 


Approximate 

relative 

afterglow 

under 

cathode ray 

excitation 


Zinc sulphide (silver) 
Zinc sulphide (silver/nickel) 

Zinc sulphide (copper) 
Zinc cadmium sulphide 

(silver/nickel) 
Zinc cadmium sulphide (copper) . 

Zinc cadmium sulphide (copper) . 
Zinc cadmium sulphide (copper) . 
Cadmium sulphide . 


Blue 
Blue 

Green 
Yellow- 
Green 
Yellow- 
Green 
Yellow 
Orange 
Red 


Medium 
Medium 

High 
High 

High 

High 

Medium 

Low 


Short 
Extremely 

short 
Very long 
Very short 

Very long 

Long 

Medium 

Short 


Calcium tungstatc 

Zinc orthosilicate ("Willemite") . 

Zinc beryllium silicate 

Cadmium phosphate . 

Cadmium borate 

Calcium phosphate . 

Calcium phosphate 


Blue 
Green 
Yellow 
Orange 

Red 

Red 
White 


Low 

High 

Medium 

Medium 

Medium 

Low 

Low 


Short 
Short 
Short 
Short 
Short 
Long 
Very long 



* R. G. Hopkinson: "Cathode Ray Tube Screens for Photographic Oscillography," 
Photographic Journal, 1946, 86B, p. 142; "An Examination of Cathode Ray Tube 
Screen Characteristics," Journ, I.E.E., 1946, 93, Part IIIA, p. 779; "Visibility of 
Cathode Ray Tube Traces in Radar Displays," Journ. LE.E., 1946, 93, Part IIIA, 
p. 795. H. W. Leverenz: "Luminescence and Tenebrescence," R.C.A. Review, 1946, 
7, p. 199. 



298 



TIME BASES 



There is another form of cathode ray tube screen in which the 
material used is potassium chloride of micro-crystalline structure. This 
type of tube, which differs from the normal only in the screen material 
employed, is known as a "Skiatron"* or dark-trace tube. Electron 
bombardment produces a deep purple-coloured trace which is due to 
a change of colour. The screen material does not fluoresce. 

The afterglow produced has a very low intensity and is best viewed 
on a translucent screen placed in front of the tube. An image of the 
picture on the tube is thrown on to this screen via a lens placed between 
it and the tube. The image is produced by shining an intense light, 
preferably green in colour, on to the face of the skiatron so that it is 
reflected from the tube screen on to the translucent screen. Obviously, 
the trace will appear black on a green or white background. 

Both heat and light are required to cause the afterglow to decay. It 
is thus necessary also to apply heat to the end of the skiatron and to 
control the degree of heat in order to adjust the decay time for the 
trace. These tubes, if kept in the dark, retain the trace indefinitely. 
Skiatrons have a considerable field of use in certain radar apparatus. 



APPENDIX I 



Table IX 



299 



Emission Peaks of Cathode Ray Tube Screens and Sensitivity 
Peaks of Emulsions 



Screens 


Colour 


Peak 
wave-length 


Emulsion 


Peak 
wave-length 


G86 
012A1 

J 

Willemite 
D 
0132A8 


Blue 
Blue 
Blue 
Green 
Green 
Yellow- 
Green 


4300 A 
44OO A 

4600 A 

5200 A 
5400 A 
5600 A 


BPi 

5B52 

5G91 
5R101 


4000 A 

4000 A 

4400 and 5700 A 

4500 and 6000 A 



The emulsion 5B52 can be developed at ioo 3 F. for specially urgent 
work. 






THE PHOTOGRAPHY OF CATHODE RAY TUBE TRACES 

The following data is given by Hercockf in connection with the 
photography of cathode ray tube traces. The papers by Hopkinson £ 
and others, referred to below r , are also of particular value. 



* P. G. R. King and J. F. Gittins: "The Skiatron or Dark-Trace Tube," Journ. 
I.E.E., 1946, 93, Part II I A, p. 822. 

f Symposium on Cathode Ray Tube Photography (Photographic Journal, 1946, 
86B): R. J. Hercock: "The Choice of Emulsion and Developers," p, 138; R. G. 
Hopkinson: "Cathode Ray Tube Screens for Photographic Oscillography," p. 142; 
W. F, Berg: "Photographic Aspects of High-speed Recording," p. 154; W. Nethercot: 
"The Photography of High-speed Traces with the Sealed-off Glass Cathode Ray 
Tube," p. 159. 

H. F. Folkerts and F. A. Richards: "Photography of Cathode Ray Tube Traces," 
R.C.A. Review, 1941. 6, p. 234; N. Hendry: "Photography of Cathode Ray Tube 
Traces," Electronic Engineering, January 1944, p. 324; W. Nethercot: "The Recording 
of High-Speed Transient Phenomena by the Hot Cathode, Glass Bulb, Cathode 
Ray Oscillograph," Electronic Engineering, February 1944, p. 369. 

X R. G. Hopkinson: "The Photography of Cathode Ray Tube Traces," Journ. 
I.E.E., 93, Part II I A, p. 808; G. F. J. Garlick, S. T. Henderson and R. Puleston: 
"Cathode Ray Tube Screens for Radar," Journ. I.E.E., 1946, 93, Part IIIA, p. 815. 



Table X 

Relative Speeds of Recording Materials 

Development: 5 minutes in 1D33 at 65°F. 





Screen colour 








Yellow- 




Blue 


Green 


Green 


5RIOI 


4 


2 


1 


5G9I 


4 


2 


2 


5B52 


1 


- 


_ 


BPi 


2 


- 


- 



In the above table unity =1-5 km./sec. at //i-o (no collimator) and 
a 1 : 1 reproduction with a beam energy of 20 uA, at 2 kV. 

The best combination for high-speed trace photography is, for most 
purposes, a blue tube with the 5G91 emulsion. 

The effective aperture is -^ — - 1 — ', where F is the aperture number 

and R is the reduction ratio, i.e. ob J ect S12e m 

image size 



3 oo 



'I 'I ME BASKS 



Table XI 



The Relative Speeds and Fogs for 5G91 Developed in various 
Developers at 65° F. 

Exposure: 20 ^A. 2 kV. Blue Screen. Writing Speed: 0*15 km. /sec. on the film 



Developer 


Time of 

development 

(min.) 


Fog 


Relative speed 

at o*i above 

fog 


1D11 


5 
10 


O-OJ 

006 


1 
2 




20 


O-II 




lD33 • 


5 
10 


009 
0-16 


3 
6 




*5 


0-27 


6 


1D47/3 .... 


1 


0-09 


2 




2 


0-12 


3 




4 


0-26 


6 



The relative speeds are obtained by working at a constant writing 
speed and varying the intensity optically. 1 Di 1 is a fine-grain developer ; 
1D33 is an active oscillograph developer; 1D47/3 is specially designed 
for rapid processing. 

The following information is taken from the paper by Hopkinson 
which has already been referred to on page 298. 

Table XII 
Exposure Times for the Repeating Traces of Television Images 





Exposure in seconds at//n 


Screen phosphor 


Fast pan. 


Fast 
ortho. 


Fast blue 


Koda- 
chrome, 
Type A 


Blue zinc sulphide 

Green willemite . 

Yellow afterglow sulphide . 

Yellow non -afterglow sulphide 

Television white 

Blue calcium tungstate 


I 

5 
3 

2-5 
2 

3 


1 

3 

3 
2 
2 

3 


I 
about 100 
about 100 
about 100 

2 

3 


10 

3° 
30 

25 
20 

30 



For a cathode ray tube running at 3 kV. on the anode, o-i u,A. per 
square centimetre beam current and using fast recording material with 



APPENDIX I 301 

an effective lens aperture //n. The table shows the exposure required 
to give an adequate record. 

Fig. 204 shows the variation of the photographic performance of some 
cathode ray tube screen materials with the anode potential. 




CR JA (Willemite) 

_ _ ,„( bellow. Zmc-cadmium-\ 
LK}0 \ $ulphicte]siiver\ I 

CR 7 (Green. Zinc -sulphide) 

CRil (Television white\ 
\ mixture / 

0, (Blue Zinc-\ 
9 \ sulphide ) 



100 



1000 
Scrcc.7 Working Potential. 



10009 Vo/ts 



FiG. 204. The variation of photographic performance of some cathode ray tube 
phosphors with the working potential for panchromatic emulsion 



Table XIII 
Maximum Writing Speeds for Recording Transient Traces 



Screen type 


Maximum writing speed (in km. 'sec.) 
on emulsion for a record of average density 


Fast pan. 


Fast ortho. 


Fast blue 


Green willemite .... 
Afterglow sulphide .... 
Blue sulphide ..... 
Non -afterglow yellow sulphide . 
Television white .... 
Calcium tungstate .... 


2 

4 

10 

4 
5 
3 


5 

5 
10 

5 

5 
3 


o-2 
01 
10 
o-i 
5 
3 






302 



TIME BASES 



Table XIII gives the maximum speed of the spot on the emulsion (not 
on the tube) for an average density record under the following conditions: 

Screen potential 3 kilovolts 

Beam current 20 uA. 

Spot diameter 1 mm. 

Effective lens aperture fz (i.e. fi at 1:1 magnification) see 

page 299. 
Recording material fast 
Developer D19B (6 minutes at 65° F.), 

For other conditions, the maximum writing speed is proportional to 
the beam current, to the square of the screen potential for the sulphides 
and to the I'Sth power of the screen potential for Willemite and 
tungstate. 

Traces with writing speeds up to one-tenth of the indicated value 
can be recorded without undue irradiation in the emulsion. With 
writing speeds of ten times the indicated value, a just visible record 
should be obtained.* 

Photographic Emulsions and Developers 

For general use: Kodak R55, Ilford 5G91, 5B51. 

Panchromatic: Kodak R60, Ilford R101. 

For utmost speed: Ilford B4235 (blue sensitive); Kodak O-Ha 

(panchromatic). 
For recording both fast and slow traces on the same film : Ilford 5B52. 
For extremely rapid processing: Ilford 5B52. 
Normal development for recording work: Ilford 1D2, Kodak D19B. 

Develop for 5 to 10 minutes at 65°F. 



The use of a Graticule 

It is sometimes convenient to photograph a wave-form super- 
imposed on a graticule. For this purpose, a thin square of celluloid or 
perspex, on which the graticule is engraved, may be mounted immed- 
iately in front of the cathode ray tube screen. When this is done, the 
best method of making the photograph is as follows : 

(a) Photograph the oscillogram with the graticule in position and 
then 

(b) Remove the time base potential and deflect the beam just off the 
screen. This will provide sufficient illumination for photograph- 
ing the graticule. 

•M. V. Scherb: "Note on the photography of Cathode Ray Tube Traces," Rev, 
Set, Inst,, 1947, 18, p. 025. 



APPENDIX I 



303 



If external illumination is employed for photographing the graticule, 
reflections from the screen often spoil the photograph. 

Errors in Photography 

Considerable errors in the apparent linearity of the trace may be 
introduced when the image on the screen of a cathode ray tube is 
photographed unless the cathode ray tube image is kept small, so that 
it does not extend beyond the relatively flat portion of the screen, and 
the camera lens is kept at a reasonably large distance from the screen. 
These precautions do not impose so great a practical disadvantage as 



/ 



X 



£NT . " . 



Centre of 
Curvature 
of Screen 



"Centre of 
Deffect/on 




Photograpft/c 
Surface 



Fig. 205. Diagram showing the method of calculating the photographic error 



appears at first sight because the resultant photographic image may be 
considerably enlarged if necessary, without introducing distortion. 

The errors which result in photographic recording have been investi- 
gated by Dodds* and by Moss and Cattanes.f Fig. 205 shows a diagram 
from which the error may be calculated. In this diagram O is the 
centre of curvature of the screen and Q is the centre of deflection of 
the cathode ray beam QP while XY is the plane of the camera lens. 
The true deflection on a hypothetical plane screen is equal to AM, 
which is proportional to tan while the deflection as seen by the lens 
is equal to BM. The error is therefore AM—BM=AB and the 

■—.-....& ».«— «^-.- ta 

• E. M. Dodds: "Recent Developments in Engine Indicators," Journal Ittst. Auto, 
Engineers, 1937, 6, p. 41. 

t H. Moss and E. Cattanes: "Errors in Photography of Cathode Ray Tube Traces," 
Electronic Engineering, 1942, 14, p. 720. 



304 TIME BASES 

combination may be found from the equation 

R- VR*-h* 



L + D 
error = — - — x 

L D+R-VR* 



# 



where R is the distance OP and k t L and D are the distances as shown 
in the figure. 

The error has the effect of crowding in the trace at the outside edges 
when the image is centrally located on the screen. 



APPENDIX II 
THE CURVATURE OF THE CHARGING CHARACTERISTIC 



When a condenser is being charged or discharged, the rate of change 
of potential across the condenser is expressed by 

dv TT i . , 

= U= volts/sec, 
dt C 

where i denotes the current in amperes, C the capacitance in farads 
and U the rate of change of potential in volts per second. 

The linearity of the saw-tooth potential wave-form across a con- 
denser is therefore conveniently expressed in terms of the variation in 
the charging current. For instance, if it is known that the current falls 
by 10 per cent during the trace then, obviously, the velocity of the 
spot will also change by 10 per cent. 

When the time base circuit consists simply of a resistance and a 
condenser connected in series across a fixed potential, together with 
some arrangement for obtaining the fly-back, the current at any instant 
during the trace is equal to the potential drop across the resistance, 
divided by the value of the resistance. The potential across the resis- 
tance is, in turn, equal to the fixed supply potential minus the potential 
across the condenser at the instant considered. If it is desired to know 
the ratio of the velocity of the spot at the end of the trace to its initial 
value, a very simple calculation is all that is required. 

Let E be the value of the fixed supply potential and let E x and E 2 

represent the initial and final values respectively of the condenser 

potential, then: 

Initial potential across R=E—E 1 

Final potential across R=E—E % . 

The ratio of the potentials, which is identical with the ratio of the 

currents, is therefore 

E—E 2 _ Final current 

E—E x Initial current 

E —E 
which may also be written as i — * -, 

E — Ei 

The amount of distortion or departure from linearity is thus indicated 
by the amplitude (E 2 — E-y) of the saw-tooth wave, divided by E— E v 

In the case of the neon tube time base circuit, Fig. 5, the potential E 2 
is the striking potential of the neon tube and E x is the extinction 
potential. To take a practical example, E % may be 150 volts and E x 
22 305 



306 



TIME BASES 



130 volts, so that V e (Fig. 6) is equal to 20 volts. If the supply potential 
E is, say, 230 volts, then: 

E z —E t 150 — 130 20 

- 71 ft — = — — 20 per cent, 

h—E 1 230 — 130 100 

and the velocity of the spot at the end of the trace is therefore 20 per 

cent less than the velocity at the commencement. 

In order to draw the actual wave-form, the equation v=E[ j_ e ~cA' ) 

may be plotted, but it must be remembered that this represents the 
charging curve of a condenser which is initially uncharged, only a 
portion of the curve being applicable to the practical case where the 
condenser is never discharged below the extinction potential of the neon 
tube or other discharge device. This wave-form is illustrated in 
Fig. 206. The curves have been plotted with the aid of the values 
given in Table XIV. 

Table XIV 



* 

CR 


t 
e c* 


t 
1 —e BS 





1 -oooo 


0-0000 


CO5 


0-9512 


0-0488 


o-i 


0-9048 


00952 


0-15 


0-8607 


0'i393 


0-2 


0-8187 


0-1813 


0-25 


07788 


0-2212 


0-3 


07408 


O-2592 


°'35 


0-7047 


0-2953 


o>4 


0*6703 


0*3297 


0'45 


0-6376 


0-3624 


0-5 


06065 


°-3935 


o-6 


0.5488 


04512 


07 


0-4966 


5034 


0-8 


°4493 


0'5507 


0-9 


0-4066 


0-5934 


1-0 


03679 


06321 


1-2 


0-3012 


0-6988 


i-6 


0-2019 


07981 


20 


<**353 


0-8647 


3-0 


00498 


0-9502 


+ 


0-0183 


0-9817 


5-o 


00067 


0-9933 


6-o 


0-00248 


099752 


8-o 


0-000335 


0-999665 


O'O 


0-000045 


0999955 






APPENDIX II 307 

Fig. 206 shows the curves for two different circuit conditions: 

In condition A E is 100 volts 

C o-i [xF. and 

R i-o megohm. 
In condition B E 1,000 volts 

C o-i tiF. and 

R io-o megohms. 

It is necessary to increase R ten times for condition B in order that 
the initial current may remain the same for both conditions. The 

E --!- 
instantaneous value of the current in the circuit is i= - e CR and is 

R 

E 

therefore initially equal to — . Since E is increased ten times, R must be 

R 

multiplied by the same factor. 






Volts 




^78! Vat t'0'2 












140 




1 3 






(A) 


„Z -- 700 voft 
CR*0-l$eco 


r 






f volts 








nd 




110 












(B)^JE~ 7000 volts 
















CR* 7-0 second 




wo 

80 
60 








A volts 








































SaaL 










40 
20 




















— 

































A current 














02 



0-4 0-6 

.t (seconds) 



0-8 



70 



Fig. 206. Exponential time base curves 



Examination of Fig. 206 will show that (for a 100 volt supply rail) 
in order to achieve the linearity obtained with 1,000 volts supply rail 
the condenser must not be allowed to charge to more than, say, 
10 volts, which is, of course, insufficient to operate the neon tube. 



308 



TIME BASES 



If, therefore, a close approximation to linearity is required, a supply 
rail potential of 1,000 volts or more is required to operate a neon tube 
whose striking potential is ioo volts. 

As may be seen in Chapters VIII and IX, efforts may be made to 
force the charging current to remain constant by a number of different 
methods, all of which aim at producing linear time bases. If the 
current can be kept constant, a perfectly linear time base potential 
will result and, in practice, perfection has been nearly achieved. There 
is a second order effect which may, possibly, be troublesome when the 
trace length occupies practically the full diameter of the screen of the 
cathode ray tube. This effect is due to the curvature of the screen and 
is discussed in Chapter VIII, page 142, and in Appendix I, page 303. 



APPENDIX III 

THE CHARACTERISTICS OF A GAS-DISCHARGE TRIODE 
USED AS A TIME BASE DISCHARGER VALVE 



The Gas- discharge Triode 

The failure of gas-discharge triode time bases to operate satisfactorily 
at frequencies appreciably above the audio range severely limits the 
value of the gas-discharge tube as a switching device. This failure is 
often erroneously attributed to the " deionization time" of the valve, 
i.e. the time interval required, after the discharge current has been 
interrupted, to allow the recombination of ions and electrons to reach a 
stage at which the grid can regain control and avoid premature reigni- 
tion. It is implied that, during this interval, the anode potential will 
be negative or zero, otherwise the term "deionization time" has no 
definite meaning since, if the anode is positive, the amount of recom- 
bination will obviously depend on the applied potential as well as on 
the time. In many gas-discharge time bases the anode potential never 
becomes zero or negative and the cessation of the discharge is not 
separated from the commencement of the charging stroke by a time 
interval which might justify the use of this term. In these time bases 
the recombination process commences during the fly-back and continues 
while the anode potential is rising, i.e. during the sweep. Thus, at 
the beginning of the trace, when the anode potential is low, the striking 
potential is also low, on account of the residual ions, and while the 
anode potential is rising (producing the trace), the striking potential 
may, or may not, return to its normal value. Even when the circuit 
does provide an interval, during which the anode is held negative, the 
observed behaviour of the time base is not fully explained by consider- 
ing the effect of residual ionization on the striking potential. Inadequate 
deionization, as revealed by the imperfect restitution of grid control, may 
reasonably be suspected when it is found that an attempt to raise the 
repetition frequency results in a marked reduction in the amplitude 
of the saw-tooth wave-form. It is, however, necessary to exercise 
caution in accepting the effect as proof of inadequate deionization 
because the presence of stray capacitance between the grid and the 
anode tends to produce the same result, since the effect is to make the 
grid less negative when the anode potential is rising more rapidly. 
Moreover, it is possible that part of this reduction in amplitude may be 
caused by an increase in the extinction potential rather than a fall in 
the striking potential. With most gas-discharge time bases, it is 
possible to reach a state where an increase of repetition frequency 

309 



3 10 



TIME BASES 



produces a noticeable decrease in amplitude, but this is not, as a rule, 
very serious. A more important factor is that, when the values of the 
components are altered to raise the repetition frequency, a point is 
reached at which the discharge fails to extinguish itself. The stability 
of the discharge is thus the dominating factor. The circuit values 
which one would select, in order to permit a more rapid rise of anode 
potential do not, in fact, make the discharge conditions unstable, 
but allow a state of stable equilibrium to be reached so that the tube 
glows steadily. In view of this, it is suggested that the deionization 
time of the tube is not directly responsible for the limitation of the 
frequency range, although it may be intimately related to the stability 
characteristics and may be found to provide a useful indication of the 
merits of different gases or electrode structures. 




OHT.+ 



R3 

r-ww* — ^ip* 



OHZ- 



Fig. 207, Basic thyratron time base circuit 

Many circuits are known in which a gas-discharge triode is em- 
ployed to discharge a condenser but, from one point of view, all these 
circuits may be conveniently divided into two classes. 

In one class the discharge is extinguished because the condenser 
behaves as a suppressor to the arc discharge, which is then only main- 
tained so long as the current exceeds a certain minimum. The other 
class, which will not be considered here, comprises all those circuits 
in which the discharge is more or less forcibly extinguished by the aid 
of additional devices which drive the anode or grid negative. The 
former class is represented by the circuit of Fig. 207, in which the 
condenser C 1 is charged through a high resistance R x and discharged 
through the triode. The resistances R 2 and R 3 limit the anode and 
grid current after the discharge has commenced. 

The Static Characteristics 

Investigations into the behaviour of this circuit have brought to light 
some interesting features which become more complex as the repeti- 
tion frequency is increased. At very low frequencies, however, the 
behaviour of the circuit is readily predictable' from the static anode 
current-anode potential characteristic of the triode. 



APPENDIX III 



3" 



Fig. 208 shows part of the anode current-anode potential curves for 
a typical argon-filled triode at various values of the grid-bias potential 
E B applied through a resistance of 50,000 ohms. The lower end of each 
curve has been omitted except in the curve for zero bias. This curve 
clearly shows a sharp change of direction where the characteristic may 
be said to change over from the hard-valve type to the arc type. The 
anode potential at the transition point is called the striking potential, 
and the valve is said to be in the ionized condition when the anode 
current is greater than that at the striking point. In order to make the 

2-0 






£ 7-0 

q 

l 



14 76 78 20 22 24 

Anode Potential (Volts) 

Fig. 208. Static anode current-anode potential curves for a typical 

argon-filled triode (GDT4B, Cossor). Grid-bias potential applied 

through a resistance of 50,000 ohms 





l\ 
























\\ 








E 3 ~0- 


--^VyJV \ 


/*""'"*' ■* 


S3 = ~ S 








£b m -?i 



measurements required to enable one to plot the anode characteristics, 
it is necessary to prevent the discharge extinguishing itself. The dis- 
charge is more easily maintained if a condenser is connected between 
the grid and the anode, the capacitance between anode and cathode 
being kept as small as possible. 

The chief effect of negative grid bias is to raise the striking potential 
by an amount which depends on the control ratio, i.e. the amplification 
factor. The striking potential is approximately equal to r times the 
grid bias, where r is the control ratio, and it thus moves considerably 
to the right of the graph as the bias is increased. One consequence of 
this, which can be seen from Fig, 208, is that an increase in negative 
bias produces an increase of anode current for a given anode potential, 
when in the ionized condition, although the anode potential increases 
if the current is held fixed. The slope also becomes less steep, indicating 
an increased value of negative resistance. 

The effect of a change in the value of the grid bias becomes less marked 



312 



TIME BASES 



as the anode current is increased, and is practically negligible from 
about 10 milliamperes upwards for the particular type of valve em- 
ployed, i.e. GDT4B.* In this region, the anode potential becomes 
almost independent of the grid bias and anode current and is approxi- 
mately equal to the ionization potential of the gas. The steep slope 
denotes a very low value of negative resistance. 

One reason why the applied grid bias has little effect, except at 
small anode currents, is that the grid current through R 3 is sufficient 



+50 




Fig. 209. 



-Z -/ 

Grid Potential (Volts) 
Static grid-current curves for a GDT4B valve 



to hold the grid potential at about — 1 volt regardless of the applied bias 
potential. This can be seen from the grid-current curves of Fig. 209. 

The curves have been taken with constant values of anode current, 
and a load line has been superimposed on the curves to represent 
# 3 =5°> 000 onms m series with a grid-bias supply of —3 volts. The 
points of intersection of the load line with the curves indicate the actual 
values of grid potential which, in this case, only fall below — 1 when 
the anode current is below 2 milliamperes. 

The choice of bias potential is, of course, governed by the striking 
potential which is desired, but the grid potential after the discharge 
has commenced is influenced to a limited extent by the value of the 
resistance R z . 

* Made by A. C. Cossor, Ltd., London. 



APPENDIX III 



313 



The Mechanism of a Gas -discharge Time Base 

The mechanism of the time base circuit as shown in Fig. 207 is best 
explained by reference to Fig. 210 which shows a number of load lines 
drawn to intersect the anode characteristic of a typical gas triode (RCA 
type 885). The condenser slowly charges through the resistance R t 
and, until ionization takes place, the anode potential of the valve is 
the same as the potential across the condenser. Assume that the time- 
constant C^g is large so that the current is maintained for a time which 
is long enough to permit the conditions to be unaffected by the rate of 




17 m \/9 W 
Anode Potential (Yoits) 



Fig, 210. Static load lines for an RCA885 gas-filled triode 



change, i.e. so that stray capacitances have no effect. Then, when 
ionization occurs, the anode current rises to a high value determined 
by the value of the resistance R z and by the difference between the 
condenser and the anode potentials. The condenser then discharges 
to a potential, say E lt by which time the anode current will have fallen 
to the value /*. In these circumstances, the potential drop across the 
anode load is 1 \R 2 , while E 1 —I 1 R 2 represents the anode potential of 
the valve. This situation is indicated by the load line R 2 corresponding 
to the condenser potential E v It will be noted that this load line 
moves to the left as the condenser discharges. Its initial position, at 
the striking point, may be imagined as parallel to that shown but at a 
considerably higher condenser potential. 

Since the time constant is large it is permissible to assume that the 
current in the valve will follow the static curve. 

Fig. 210 also shows a load line marked 7? 1 +i? 2 which is drawn in 



3*4 TIME BASES 

such a position that the charging and discharging currents are equal 
and hence the potential on the condenser will not alter. This means, 
in effect, that the discharge through the valve cannot be extinguished 
in the condition postulated. 

In order to enable the discharge current to fall to zero it is essential 
for the (i? x +R 2 ) load line to strike the curve at a point below that at 
which the R 2 load line becomes tangential to the curve. In this case, 
when the current has fallen to the tangential point it cannot follow the 
curve any farther since this requires an increase in the potential across 
the condenser, in spite of the fact that the charging current is less 
than the discharging current. Therefore, since the condenser potential 
cannot rise, the discharge is suddenly extinguished, leaving the con- 
denser charged to the potential indicated by E 2 . It may therefore be 
said that extinction occurs at the tangential point where the negative 
slope resistance of the arc characteristic becomes equal to the positive 
resistance R 2 . The diagram of Fig. 210 has been plotted for an 885 
valve with R 2 equal to 500 ohms, R 3 equal to 50,000 ohms and a grid 
bias potential of —6 volts. 

The above theory has been checked by first making R x too small to 
permit extinction of the discharge current and then gradually increasing 
it and noting the current at which oscillation commences. With con- 
densers of the order of o-i microfarad or larger, reasonable agreement 
is found between the measured current and the value predicted from 
the static characteristics. This check was made on several valves and, 
in each case, the resistance R 2 was varied from 500 to 5,000 ohms. If R 2 
is made considerably higher, the discharge is apt to extinguish itself at 
higher currents than would be expected, and it is found that the stray 
capacitance between anode and cathode is then able to extinguish the 
discharge before C x can do so. Normally, R z is only made high 
enough to limit the peak current to a safe value, of the order of 200 
to 500 milliamperes. 

So far we have only considered the case where C t is rather large, 
of the order of o-i microfarad. When a considerably smaller capacitance 
is used, it is found that the charging current has to be reduced in order 
to maintain oscillation. In other words, the real extinction point falls 
below the tangential point on the static characteristic, which now no 
longer provides an indication of the existence of the unstable conditions 
which make the discharge self-extinguishing. An example of this is 
given in Table XV. These results were obtained with the circuit of 
Fig. 207 using a GDT4B valve with -4.5 volts grid bias; R 2 was 500 
ohms, while i? 3 was 50,000 ohms. The second column gives the 
minimum current which can be continuously maintained without 
oscillation when R x is very slowly increased. 



APPENDIX III 
Table XV 



3 J 5 



Capacitance of Cj 


Extinction current 


!*F. 


mA. 


4-0 


1-84 


o*i 


1-83 


0-02 


i-8o 


0-002 


i-6o 


0-001 


1-44 


0'0005 


1-04 


O-0O02 


o-6i 


o-oooi 


0-38 


O 


0*09 






It will be seen that the extinction current begins to fall almost in 
proportion to the capacitance when very small condensers are used. 
Thus the frequency of a gas-discharge time base cannot be indefinitely 
raised by reducing the capacitance, because the charging current also 
has to be reduced. 

While it is not suggested that this extinction current, determined 
by static means, is identical with the extinction point when the triode 
is actually operating as a time base, it does signify that a time base 
which employs a higher charging current than this will not commence 
to oscillate until it is given a shock. It is actually found that the 
charging current can be only slightly increased above the extinction 
point as shown in Table XV without stopping the oscillation completely. 

The existence of a stable equilibrium condition below the tangential 
point implies that the condenser potential rises when the charging 
current through R t is momentarily reduced below the value of the 
discharge current through R 2 . The explanation of this paradox is that 
the condenser is shunted by a negative resistance and the ordinary 
course of events is therefore reversed. 

When the condenser is small, the variation of potential from the 
junction of R x and R 2 to earth is the same as would be expected without 
a condenser except that the discharge still extinguishes itself at a point 
farther along the curve. In other words, the small condenser only 
takes control of the discharge when the current has fallen well below 
the tangential point. We may then consider that the dynamic slope of 
the discharge characteristic has become equal to the slope of the /? 8 
load line at the actual extinction point. 

It will thus be seen that, for small values of capacitance, the charging 
current must be considerably reduced and for this reason, apart from 



316 



TIME BASES 



any others, the maximum frequency of a time base of this type is 
limited. 

One reason why the dynamic slope is steeper than the slope of the 
static characteristic is the presence of stray capacitance between grid 
and anode. This will cause the grid potential to rise in a positive 
direction when the anode potential is increasing rapidly. It will be 
seen from Fig. 208 that this will result in a reduction of anode current 
and is equivalent to a steepening of the slope. 

The effect of this stray capacitance can be enhanced by adding 
further capacitance across the grid and anode of the valve. This is 




0-ooot 



000/ 001 

Capacitance ((/.f) 



Of 



Fig. 211. Characteristic curves showing the minimum current for stable equilibrium 

with various values of added capacitance between anode and grid for a GDT4B valve 

(i? 2 = 500 ohms and i? 3 «■ 50,000 ohms) 

illustrated in Fig. 211 from which it will be seen that, for small values 
of the main condenser C lt the anode current at the extinction point 
falls to a greater degree as the additional grid-anode capacitance is 
increased. The curve for Opf is, of course, plotted from Table XV. 
It is perhaps reasonable to assume that, in the absence of any capacitance 
between anode and grid (including that between the anode and the 
cloud of ions), the extinction current would remain constant, in this 
particular case, at 1-84 milliamperes. On the other hand, the known 
existence of hysteresis loops in the characteristics of arc discharges can be 
satisfactorily explained by saying that an increase of current is accom- 
panied by an increase in the number of ions and that the production 
of the additional ions requires energy, which is obtained by a temporary 
rise in the anode potential.* 

• W. G. Dow: Fundamentals of Engineering Electronics, J. Wiley and Sons, Inc. 
(Chapman and Hall, Ltd.). ' 



APPENDIX III 



317 



From the foregoing remarks it would appear to be reasonably certain 
that the action of the main condenser C\ in extinguishing the discharge 
is similar to the effect of a condenser in extinguishing an arc across a 
pair of contacts. 

Some examples of the behaviour of a gas-discharge time base were 
obtained by connecting an X plate of an oscillograph to the anode of 
the discharge valve, while a Y plate was connected to indicate the 




-GH.I+ 






300 V 



Lo^r- 



Fig. 212. Circuit for examining the behaviour of a gas-filled triode time base 

potential drop across R 2 as shown in Fig. 212, in which the components 
have the following values : 



*1 




i-o megohm (max.) 


R* 




500 or 5,000 ohms 


R$ 




50,000 ohms 


R 4 


and R b 


1,000 ohms 


v 1 




RCA valve type 885 






These cyclograms are shown in Fig. 213. As the charging current is 
very much smaller than the discharge current, the potential across R 2 
is, except for the effects of stray capacitance and inductance, approxi- 
mately proportional to the anode current only. From the curves it 
will be noticed that the potential developed across R 2 appears slightly 
less than the potential through which the anode has fallen. This is 
due to the potentiometer effect produced by the presence of resistance 
and inductance in the condenser C x and its leads, together with that 
produced by the resistance R & and the stray capacitance from the Y plate 
to earth. 

When the capacitance of C x is large, as in cyclograms (1) and (2), the 
discharge clearly follows the static characteristic down to the anticipated 
extinction point and then the current falls very rapidly, as is made 
evident by the fact that the trace becomes very thin, to zero. 

The peak value of the discharge current is about 150 m A. in (1) and 
15 mA. in (2). The zigzag effect at the higher current is believed to 



3i8 



TIME BASES 



VPPENDIX III 



319 



be due to the glow jumping from point to point. This is often seen 
when plotting static characteristics at high currents. 

During the rapid changes of current at the beginning and end of 
the discharge, the spot moves across the screen at an angle of approxi- 
mately 45 because the condenser potential remains constant during 
these short intervals and the X and Y plates, which are connected 
directly across the condenser, therefore undergo equal potential changes. 
The angle will be precisely 45 ° only if the deflection sensitivity of 
the X and Y plates are equal. 

When the capacitance of C x is reduced, the locus shows a tendency 
to overshoot or break away from the static characteristic and the anode 
potential appears to reach a minimum in the region of zero for a short 
time. 

Sometimes the anode voltage does not merely touch this minimum 
as shown in (5), (6), (7), (8) and (10), but remains there for a definite 
short time. This is clearly indicated in cyclograms (4) and (9) by the 
spot which indicates a long-term residence at zero anode potential. 

It has been found that the duration of this low-potential regime 
may be increased to as much as 20 to 30 microseconds by connecting 
an additional condenser directly between anode and cathode. The 
occurrence of this regime cannot be readily detected by a change in 
the appearance of the glow unless the duration of the low anode 
potential regime is many times longer than the duration of the ionization 
potential regime. A change in the colour and luminosity of the glow 
can then be easily observed. The normal glow obtained with argon, 
which may be described as magenta in colour, is replaced by a pale blue 
glow and the discharge becomes much less brilliant. 

The low-potential regime is found to exist in all the diagrams from 
Nos. (4) to (10) inclusive and it is believed that it also exists in the 
first three diagrams, but that, since the repetition frequency is so much 
lower, it is not visible on the tube. For the same reason, the trace 
from the maximum value of potential across the X plate to the maxi- 
mum value of potential across the Y plate is hardly visible in any of 
the diagrams from Nos. (1) to (6) inclusive and for this reason they 
have been shown dashed. Examination of the diagrams (4), (6) and 
(8) shows the low-potential regime continuing to exist for progressively 
longer portions of the cyclogram. This, however, does not mean that 
the regime lasts for a longer time, since the oscillograms are also 
progressively more rapid and it is, in fact, probable that the time taken 
for the potential to change from the neighbourhood of zero to the 
ionizing potential is more or less constant. 

If the capacitance is still further reduced below the value of o-oooi 
microfarad as shown in diagrams (9) and (io), it will be found that it is 





R 2 = 500ti> 


^ 2 " 5000 Li 


Wis 








1 (1) 










ay 


75 




%• 


\ 

\ 


f=40c/s 






! 


\ 


Cf-ClpF 
f=40c/s 


50 
25 






















\ 








l 


^ 


\ 
\ 


s 






V 


^ 


\ 
\ 


\ 

s 

V 

































(3) 




V 






(4) 


75 

SO 
25 




r 


\ 


C t =002fiF 

f=200c(s 




x 




^ 
_^__ 


f=200cfs 




< 


> 


\ 










\ 

\ 
\ 


\ 
\ 






\ 


V 


\ 

v 


s 






l 


^ 




\ 

N 






















I 75 

X50 

,"> 
%25 

Q 
« 

!a 
C 










(5) 










(6) 




; 




CrO-OOSpF 
f=8O0c/s 




s 


\ 


Ci=OMIfiF 
f=4kc 




/ 


\ 


1 






\ 


\ 


\ 






(j 


V 


\ 
\ 


1, 




X 


V 


\ 


\ 
\ 






















J" 

§75 

50 










(7) 










(8) 




r 




Ci=O-0O!pF 






s 


d=000Q5^F 
f=8kc/s 














r 


> 


\ 

s 




'15 






I 


V 








\ 


v^ 


\ 

\ 


\ 






















75 

50 

25 












m 










(10) 








C^OWOIpF 
f~40kc/s 




ri 


s 


&*0'006tftF 
f=40Ac/s 














t 












V 


N 


\ 




r 


V 





















25 50 75 25 50 75 i/o/ts 

Anode Potential ' (XdeF/ect/on) 

Fig. 213. Time base cyclograms obtained with the circuit of Fig. 212 



320 



TIME BASES 



impossible to charge the condenser to the same striking potential across 
the X deflector plates. This occurs because a certain amount of time is 
required for the recombination of the remaining ions after the discharge 
is extinguished. Since the capacitance has been reduced, however, the 
time available for recombination (or deionization) is reduced because 
the condenser charges more rapidly. The striking potential, in the 
presence of a few ions, is lower than when no ions are present and, 
for this reason, the discharge is initiated before the small condenser 
has had an opportunity of charging to a potential which is as high 
as that which would be required for a larger condenser. This condition 
exists only when the condenser is so small that the circuit approaches 
the value for non-extinction of the discharge and the condition is not, 







L 









I LyywJ 

■?■ Co 



^WVV — rOMZ+ 



\Cj 300 V. 



*-Gft.T- 



FlG. 214, Circuit for examining the behaviour of a time base in which 
the anode load resistance is replaced by an inductance 



therefore, a very important factor in determining the limitations of the 
circuit. 

The behaviour of the time base circuit when R 2 is replaced by an 
inductance can be observed by employing the circuit of Fig. 214. 
In this circuit, the anode potential is applied to an X plate of the 
oscillograph while a Y plate is connected to the condenser. 

Cyclograms obtained with this circuit are given in Fig. 215, from 
which it will be seen that the discharge continues until the charge on 
the condenser is reversed in direction. The inductance is then set into 
a damped oscillation with its own self-capacitance plus the stray 
capacitance C of the anode. The stray capacitance coupling between 
anode and grid transfers this oscillation to the grid as seen in the right- 
hand cyclogram which shows the relationship between the grid potential 
and that across the condenser C v It should be noted, however, that 
this oscillation does not appear across the condenser, and the resultant 
time base trace is therefore not affected, but a damping resistance i? 6 
may be added across the inductance L if desired. 

These results were obtained with the following components for the 
circuit of Fig. 214: 



APPENDIX III 



321 



R t and R s 
L 



0-5 megohm 
50,000 ohms 
1,000 ohms 
0*02 txF. 
17 mH. 



It will be noticed that a spot exists at the bottom of the cyclogram 
in the region of zero anode potential. This is apparently due to the 
fact that the inductance attempts to drive the anode negative and in 
so doing causes it to collect the ions which still remain after the anode 
potential has become too low to maintain ionization. As a result, the 
potential is held approximately at zero until most of the remaining 
ions are removed by the anode and grid. 

Volts 
+100 



+ 75 

I 

*+25 






-25 



-50 

















y 1 ^ 






y 












O 










































/kJ 



































t 


i 







-75 -50 -25 



+25 . +50 
A/rode Potential 



+75 



HOO -25 Volts 
Grid Potential 



Fig. 215. Cyclograms obtained with the circuit of Fig. 214 



It is also of interest to note that the grid potential is held at approxi- 
mately zero during the discharge period, including the short interval 
at the end, where the anode potential pauses in the region of zero. 

Inspection of the cyclogram (Fig. 215) will reveal the existence of the 
low- potential regime (loop at top of discharge) at the commencement 
of the discharge exactly corresponding to that obtained when using 
the circuit of Fig. 212. Furthermore, an increase of the value of the 
capacitance C causes the duration of the low-potential regime to 
increase in this case also. 

It is thought that the low anode potential regime at the commence- 
ment of the discharge may be brought about as follows. During the 
23 



322 



TIME BASES 



initiation of the discharge the anode potential falls very rapidly and 
the grid is driven negative by stray capacitance to such an extent that 
the anode current-anode potential locus rises as shown in the cyclo- 
grams of Fig. 213 instead of tracing out the static characteristic. The 
resultant concentration of ions in the neighbourhood of the anode, at 
the moment when the anode reaches the ionization potential, may 
therefore appreciably exceed the concentration under steady conditions. 
If this is the case, the random electron current density in the ionized 
gas will also be higher than normal and the anode will therefore be 
able to collect an appreciable electron current for a short time while 
the anode potential is forced well below the ionization potential. 

In conclusion, it can be stated that, at least for time bases of the 
type shown in Fig. 207 using argon-filled valves, the original idea that 
denization time is alone responsible for the failure of the time base 
to operate at frequencies greater than about 40 kilocycles is erroneous. 
The limitation is actually due to the necessity for reducing the charging 
current as the capacitance is reduced, in order to obtain extinction of 
the discharge. If extinction can be obtained by other means, deniza- 
tion time may be expected to become a limiting factor. 

It should be noted that deionization time depends upon the circuit 
conditions as well as upon the valve, since the time required for ionic 
recombination depends upon the electrode potentials. 

The traditional explanation of the process of ionization appears to 
be incomplete in that evidence has been brought to light which shows 
that the anode potential falls to the neighbourhood of zero before 
becoming stabilized at the ionization potential. 






APPENDIX IV 
DIFFERENTIATING AND INTEGRATING NETWORKS 

Much confusion is caused by the employment of verbal expressions 
which are not precise, and the terms "Differentiating circuit" and 
"Integrating circuit" generally fall within this category.* Sometimes 
these terms are used to denote a circuit which is so designed that the 
output potential is accurately proportional to the time derivative or 
integral of the input potential so long as the input potential falls within 
certain predetermined limits of wave-form, frequency and amplitude, 
but the names are more commonly associated with the potential- 
operated networks of Fig. 216 and the current-operated networks of 
Fig. 220. This appendix is mainly concerned with the properties 
of these elementary circuits, the applications of which are many and 
varied. The possibility of obtaining a reasonably close approximation 
to true differentiation and integration, in the mathematical sense, is 
very often a secondary consideration; the primary function of these 
devices is to develop or subdue some characteristic feature of the 
wave-form applied to the input terminals. The examples given in 
Figs. 217, 218 and 219 illustrate this behaviour, and it will be noted 
that the differentiating networks tend to enhance the sudden changes 
of the applied wave-form while the integrating networks tend to smooth 
them out. The circuits shown in Fig. 220 produce results which are 
similar to those obtained with the circuits of Fig. 216. 

Fortunately, it is seldom required that a circuit shall differentiate 
or integrate accurately and, indeed, especially in the case of differen- 
tiation, the ability to discriminate between slow and rapid changes is 
generally all that is required. For integration, a closer approach to 
the genuine integrating condition is usually required, but it is not 
impossible to obtain the required result at the expense of a certain 
amount of complication (see Chapter IX and also page 340). These re- 
laxed requirements are a particularly important factor, since the simple 
forms of so-called differentiating and integrating networks only 
provide an output which closely resembles the true mathematical 
result when that output is a negligibly small fraction of the input 
potential. 

When a pure sinusoid is applied to a differentiating or integrating 

network, the wave-form of the output is not changed. Since there is 

only one frequency present in the source, no change of form can take 

* N. Marchand: "The Response of Electrical Networks to Non-sinusoidal Periodic 
Waves," Proc. I.R.E., 1941, 29, p. 330; F. C. Williams: "Introduction to Circuit 
Techniques for Radiolocation," Joitrn. I.E.E., 1946, 93, Part IIIA, p. 389. 

323 



324 



TIME BASES 






place. Such a change of wave-form would imply an alteration in the 
harmonic content of the wave and a passive circuit cannot effect this change 
if the applied wave-form contains no harmonics. The only result which 
occurs is that the output differs from the input in amplitude and phase. 
An example will serve to clarify the situation. 

Mathematically, the differentiation or integration of a sine wave 
produces a cosine wave: the input and output voltages should therefore 
differ in phase by 90 . With a simple circuit consisting only of two 
elements, such as a condenser and a resistance, the actual phase dif- 
ference between input and output, with a sinusoidal wave-form, is not 
90° under practical conditions. If the output potential amplitude is 
one-half that of the input the phase shift will be only 6o°. If the 
output is 5 per cent of the input, the phase shift becomes 87 , but 
with such a large loss of amplitude it is generally necessary to employ 
an amplifier. 

The ratio of output to input potential for a sinusoidal wave-form 
is given by 

for a simple differentiating circuit, 



Vl+a>*C*R* 



for a simple integrating circuit, 



pCR 
1 +pCR 



and — 

where o> = 2r/ (/ =frequency) and CR is the time constant 
These may be written in Heaviside operational form as 

for a simple differentiating circuit, and * _ - for a simple integrating 

circuit, where p is the Heaviside operator. 

If the inductive form of circuit is employed, the time constant is L/R, 
and this is used in the above expressions in place of CR. 

Simple Forms of Differentiating and Integrating Network 

Differentiating and integrating networks may be energized by a 
potential or by a current and the type of network to be employed is 
different for the two cases. When any network or circuit is fed from 
a source whose internal impedance is low compared with that of the 
network, the input is in the form of a potential. On the other hand, 
when the source impedance is high compared with that of the network, 
as, for example, when the network is inserted directly into the anode 
circuit of a pentode valve, or when it is fed from a pentode whose anode 
load resistance is of a high value, the valve passes a current ge into the 
anode load resistance and network in parallel. Since, however, the 



gt 




* 1 x 



at f 

Mi 

2 P 



$ 




■- 
-^ 

tr. 

E 
I 

3S 
i 







3 
5 



s I? 






3 H 
o 

i 




& is s 4 s> s s 

j r— r — i — T — | 1 r 





i- c 

P * 

10 > 

-J W 



*-*-MS 



1 J 


l 4 


gi&oisfessi 




s 




t 


* 




i 

-i 
i 


5 




i 


f» 




i 






I 5 


- 




> 


- 




> 






** 






n. 






<i 








* 


' 



3 ' v 




| 

n 



:&j 



- r 555S?F l - 



* o 



t ° 



to 

Hwwv-f 



g> 



ja 5 



^ 


rf 


s 


| 


PS 


ei 


o 


^ P 


z 


*s 


en 




H 


-i 


> 


> 



-vwvw- 



i 



N 



-II- 



I * 






APPENDIX IV 



325 



anode load resistance is high compared with that of the network 
substantially the whole of the current passes into the network. It will, 
therefore, be obvious that high impedance networks must be employed 
for potential operation and low impedance networks for current 
operation. 

Fig. 216 shows a series of high impedance networks, of which two 
provide a differentiated output and two give an integrated output 
potential. Fig. 220 (page 327) shows a series of low impedance 
networks. 

It is of importance to note that the impedance of any apparatus 
connected to the output terminals of either the high or low impedance 
types of differentiating and integrating networks should either be so 
high that any current drawn from the network is negligible compared 
with that flowing in the network, or the load impedance should be of 
the same type as that of the shunt arm of the network across which it 
is connected. In this case, the impedance of the load must be taken 
into account in calculating the value of the shunt arm and that of the 
time constant of the network. 



High Impedance Networks 

As already stated, Fig. 216 shows the details of practical high 
impedance differentiating and integrating networks. Fig. 217 shows 
the output wave-forms obtained for a square-wave potential input for 
various values of the time constant CR or LjR. In these curves, and 
in those which follow in Figs. 218 and 219, the statement CR=$T, for 
example, means that the time constant CR or LjR has a value which is 
five times that of the duration of the linear portion T of the input wave- 
form. 

Fig. 218 shows the output wave-forms for the same networks when 
the input potential is of saw-tooth wave-form and the fly-back time is 
zero. It should be particularly observed that, in the case of the inte- 
grating networks, the parabolic output wave-form is somewhat "lop- 
sided," especially for the smaller values of the time constant CR or LjR. 
This lopsided condition is made evident by the fact that, although the 
crests reside at times t—o and * = i, the roots appear at times £=0-32 
and ( = 1-32 when the time constant is 0-2?'. 

Fig. 219 shows the output wave-form when the input potential is 
of triangular form. 

Examination of all the output wave-form curves will demonstrate 
the fact which has already been stated that perfect differentiation and 
integration cannot be obtained by means of simple circuits. Reference 
to the output wave-forms obtained from differentiating networks will 



326 



TIME BASES 



show that the time constant of the network must be very small compared 
with the duration of the linear portion of the input potential and that 
the input and output wave-forms differ only slightly if the time constant 
is equal to the time T. Moreover, in the case of integrating networks, 
the time constant must be large compared with T if the output wave- 
form is to resemble the integrated form of the input wave-form. 
Unfortunately, these two conditions act to restrict the output potential 
available from the networks. Actually, perfect differentiation and 
integration are only obtained when the output potential is zero, hence, 
a compromise is necessary. For the resistance-capacitance type of 

differentiator, the output potential is approximately C7?-? and for the 



integrator it is approximately j^\Edt volts. 



dt 



For the resistance-inductance type CR is replaced by LjR in the 
above formulae. 

It should be noted that the output wave-form is, in every case, that 
which appears across the shunt arm of the network. Moreover, the 
output wave-form for a differentiating circuit is the same as that 
appearing across the series arm of an integrating circuit, and vice versa, 
provided that the input conditions and the time constants of the two 
circuits are identical. Theoretically, it is of no consequence whether 
the circuit is composed of inductance and resistance or resistance and 
capacitance but, in practice, it is inadvisable to employ an inductive 
differentiating circuit, if it can be avoided, because the effective resis- 
tance of the coil introduces an output component in addition to that 
due to the inductance. If that component of the output potential 
which is due to the inductance is only a small fraction of the input 
potential, as is usually the case, the potential developed across the 
resistance of the coil will have practically the same wave-form as the 
input potential. The ratio of the amplitude of this potential to that 
of the input potential will be almost the same as the ratio of the resis- 
tances. The practical importance of this effect depends very much on 
individual circumstances. 

In this connection, it is to be noted that in using a capacitive 
differentiating circuit, the series capacitance should preferably be large 
compared with the input capacitance of any valve, or other device 
connected across the shunt resistance, in order to avoid increasing 
the time constant by an indefinite amount, but no distortion of the 
type encountered with the inductive differentiating circuit will normally 
arise. 

When a current is passed through a resistance in series with a 
condenser, the rate of change of the potential E e across the condenser, 






APPENDIX IV 



327 



at any given instant, is proportional to the instantaneous value of the 
current. Moreover, the instantaneous value of the current is propor- 
tional to the instantaneous value of the potential E r across the resistance 
so that: 

(a) E T is proportional to the rate of change, or time derivative, of E c , 

(b) E c is proportional to the time integral of E r . 

The potential applied to the condenser and resistance in series is 
equal, at every instant, to the sum of the potentials E r and E c . Then, 
if E c is very small compared with E r} it follows that E T must be approxi- 
mately equal to the applied potential. Therefore: 

(c) E c is approximately proportional to the time ititegral of the applied 

E 
potential when -* is small. 

Similarly, 

(d) E f is approximately proportional to the time derivative of the 

E 
applied potential when -- is small. 

c 

In applying these rules, it is to be understood that the con- 
denser must be initially uncharged; otherwise the rules will not 
apply until sufficient time has elapsed to allow the initial charge to leak 
away. 

Low Impedance Networks 

As already stated, Fig. 220 shows the details of low impedance 
differentiating and integrating networks of which one, Fig. 220 (a), is an 
ideal differentiating network 
and Fig. 220 (b) and (c) are 
practical networks. Fig. 220 (d) 
is an ideal integrating network 
and Fig. 220 (e) is a practical 
integrating network. The cir- 
cuits of Fig, 220 (a) and (d) are 
ideal and not practically realiz- 
able since pure capacitances 
and inductances are not them- 
selves physically realizable so 
that, in practice, one is forced 
to employ the networks of 
Fig. 220 (b) and (e). The time 
constant of each of the ideal 
networks is infinity since, in the 



I 



(cv) 





\C $R e 



(e> 



Differentiating 
Circuits 



Integrating 
Circuits 



Fig. 220. Simple low impedance differentiating 

and integrating circuits requiring a current 

input 



328 



TIME BASES 



case of Fig. 220 (a) the value of the resistive component of the network 
is zero and, in the case of Fig. 220 (d), it is infinite. Fig. 220 (c) is 
also physically unrealizable, nevertheless transformers are often em- 
ployed for differentiating purposes in those cases where the losses in 
the transformer may be neglected. In the ideal network shown in Fig, 
220 (c), the time constant is infinite since the circuit resistance is zero. 
In the practical form of the circuit, the resistance component of the 
network is finite, and it is also modified by the presence of the output 
load. For these reasons, only the networks shown in Fig. 220 (b), (c) and 
(e) need be considered. 

For these networks, the time constant is CR or LjR, according to the 
circuit component employed. 

These low impedance networks must be fed from a source of current 
and, for this reason, they are usually placed in the anode circuit of a 
pentode or beam- tetrode valve. 

The networks shown in Figs. 216 and 220 and others of a more 
complicated nature, are also known as linear shaping networks. 

Response of Common Types of Simple Differentiating and 
Integrating Networks to Sine Waves and to Step Functions 
of Amplitude and Velocity 

It is of the greatest importance to be able to determine rapidly the 
response of the foregoing networks to wave-forms of various kinds and, 
for this purpose, the author cannot do better than to quote Williams* 
in his own words. 

The process of linear shaping marks the first major difference 
between the analytical approach to circuits of the type under discussion 
and that appropriate to communication circuits where wave-forms are 
either sinusoidal or can be treated as an aggregate of sinusoids 
lying within a certain frequency band. This will be clear at once by 
reference to Figs. 221 and 222, which show the result of passing, first 
a sine wave, and then a square wave, through two simple RC 
networks. 

With a sinusoid the networks introduce a change of amplitude and 
phase, but no change of shape. With a square wave both networks 
introduce a vast change of shape, and phase has become meaningless, 
its place having been taken to some extent by a correspondence in time 
between discontinuities in input and output waves; this correspondence, 
however, differs vastly from the conception of phase, since a time shift 
is not involved. While it is still theoretically possible to use the 
sinusoidal component approach by splitting the input wave into its 

, *J' C - W H 'L an 2f : " Intr °duction to Circuit Techniques for Radiolocation," Journ. 
LE.E., 1946, 93, Part UFA, p. 289. J 



Input 



APPENDIX IV 

Output 



Input .. 
- — II- 






ROT 



I 



3 2 9 
Output 



CR«T 





Fig. 221 



Fig. 222 




EJcponential to lero of 
time constant RC 



Fig. 223 



33° 



TIME BASES 



Fourier components, handling these by the normal methods and sub- 
sequently recombining them to find the output, this process would be 
laborious in the extreme and is quite impracticable. Its place must be 
taken by a much less formalized approach which is largely intuitive and 
which rests heavily on previous experience and on step-by-step physical 
argument. 

To illustrate this by a very simple example, consider the circuit of 
Fig. 222 with a square wave applied to it, and suppose that the square 
wave has just started, as indicated in Fig. 223. We know that if the 




Fig. 224 

starting level of the input potential has been maintained for a 
long time the potential across R will be zero; hence we can draw 
the output potential curve (b) up to the first rise of the square 
wave. Furthermore, since a finite current cannot change the potential 
across a condenser instantaneously, the whole of this rise will appear 
instantaneously across R, so that curve (b) rises as indicated. So far, 
the circuit constants have had no effect on the result, but the next phase, 
in which we have current flowing in R and producing a change in the 
potential across C, will depend, for its time scale, on the product RC. 
In fact, the next stage is an exponential decay of output potential of value 
Ye-VRC^ w here t is measured from the instant rise of input potential. 
This can be drawn in curve (b) up to the point on the time axis when the 
next downward discontinuity occurs in the square wave. Considering 






APPENDIX IV 



33i 



first the case where RC is small compared with T, the decay will be 
sensibly complete by this time and the next step can be taken as indicated. 
This step will be subject to the same starting conditions as the first and 
the calculation process is complete. If, however, the second discon- 
tinuity in the input occurs before the disturbance created by the first 
one has died away, new starting conditions apply to the second dis- 
continuity (see curve (c)) and the calculation becomes more difficult. 
If the number of steps is limited, e.g. when the input is a short pulse 
(see Fig. 224, curve (a)), an alternative method involving the principle 
of superposition is valuable. The pulse can be regarded as the sum of 



(a) 































V 

1 














- 


VI 


1 










1 


r. 


V 






f 



(b) 



Fig. 225 

two wave-forms, a positive step, curve (b), and an equal negative step 
delayed a time T, equal to the pulse width, after the positive step, see 
curve (c). The principle of superposition states that the response to 
the pulse is equal to the sum of the responses to these two components. 
The components' responses are shown by curves (d) and (e) for RC > T. 
The total response, (/), is the sum of these. If the pulse is isolated, 
or repeated at intervals much greater than RC, this approach is 
sound. 

If the input is a square wave with T ^> RC } either of the above 
methods can be used, provided that in the first method the new 
starting conditions for each discontinuity are taken into account 
and the process is continued until it begins to repeat itself, and 
provided that in the second method a sufficient number of com- 
ponent step functions is considered, so that the disturbance due to the 
first has sensibly ceased when the last one occurs. Both methods are 
laborious and often a quicker solution can be found by an appeal to 
principles of symmetry. Thus the answer must clearly be symmetrical 
about earth and have the general shape shown in Fig. 225, curve (b). 



332 



TIME BASES 



If we let v be the amount by which the exponential fall fails to reach 
earth, the initial value will be V— v hence 

v =(V~ v)e~ TIRC 



or 



o = 



y -Tine 



and the output is known. 

The first of these methods of approach is by far the most useful, 
because in the vast bulk of circuit analysis, either the transient has died 
away before the subsequent discontinuity occurs in the input wave- 
form, or else some non-linearity in the circuit is encountered which 
effectively resets the starting conditions for the next phase of operation 
to new known values which are not dependent on the extent to which 
previous disturbances have died away. 

Aids to Rapid Determination of the Shaping Effect of a Network 

To use the above-mentioned first method of approach it is necessary 
to be able to arrive quickly at the change of shape imposed on a given 
input wave by a given shaping network. Clearly it is impossible to 
memorize the changes imposed on all wave-forms by all networks, and 
some simplification is essential. Fortunately, nearly all the useful 
wave-forms can be broken down into a series of component step 
functions of either amplitude or velocity; e.g. the square wave is a 
series of time-displaced equal and opposite amplitude step functions; 
the start of a linear time-base is a velocity step function, and so on! 
This reduces to two, the common input wave-forms that need be 
memorized. 

With regard to the networks, it is no longer permissible to calculate 
impedances and transfer functions in terms of R, j<*L and i/(/o>C), 
since such calculation would relate only to the component of input of 
frequency o>/(2*). It is quite necessary, however, to be able to write 
down some mathematical statement of the network and to have names 
for the equivalent of transfer function and impedance. The practice 
that has been adopted is to calculate these quantities in the normal 
manner but with <j>=d(dt) y written for>,* and to continue to refer to 
the quantities as impedances when referring to them in circuit dia- 
grams, but to refer to the mathematical representation as an operator. 
Thus stated, the quantity is used as a label for a remembered change of 
shape rather than as a shorthand approach to the differential equation 
of the network; its usefulness in this connection is demonstrated in 
Table XVI, which shows in tabular form the operators of the 

Oxford UnW^'pTet/r^. 1 ^" ° perational Methods-Applied Mathematics, 







si. A_ Lt££5 



cm to At 



$+$ /?,+/?2 



ft 



**A\r^v$ 




Initial rate 




M: 






—i^-i-O > 



■elocitv 



HA 



[To face page 332 



N 









-f-e,-e 



mf Of 



R^ C±(e 
/{input) 



*\ ■■ fl 



T 



i 



ft _J?_ 






*. joZ 



* F$2M 






I Initial rate-A 



t-0 



e=0 OF 

'k(t<R) 




t-o 



e-0 



♦pCP 



/?/ 



r MCta£'f3 






t-0 



^ o- 

Initial rate e z 

CRtokCR 






Initial rate 
./ff , 
. Z ^toZ* 



Table XVT 



E 


a | & 




See cases El and N. with appropriate 
scale changes 


71 


■^TTT 1 pi 




See case I with appropriate scale 
change 


13 


If 

T! 


f pC 


t-0 

t-0 f-0 


2d 


L 


i 1 


See case ED with appropriate scale 
change 


B 


3 


i.l.-L 


See case 1 with appropriate scale 
change 



CaseX 



R3 c 



T JL.JL l^pffi 



ti 



' u fl 



I 



Xw *J% X * 





Initial rate 4: 
0?z 









Response of common differentiating and integrating circuits to step (unit) functions of amplitude and Telocity 



[To face [><i«e ;,^z 









APPENDIX IV 



333 



commoner networks, together with the responses to step functions of 
amplitude and velocity. Of the networks shown, the only one for which 
a mathematical solution need be sought is the first; the remainder 
become obvious by inspection if the operator is broken into its com- 
ponent parts and applied step by step in the appropriate order. It is 
quite essential to memorize this first case, not only because it occurs 
extremely frequently but also because so many others rest on it, as may 
be seen from the table. Its properties may be summarized by the 
statement that, whenever it is disturbed by a discontinuity in the input, 
it will approach a steady state, which is usually obvious, exponentially 
with time-constant T=CR. Note, in this connection, the method by 
which this is indicated in the figure, by the arrow and the annotation 
"CR to E" which is shorthand for "exponential approach to E with 
time constant CR." 

Turning now to case (ii) for which eji—Rj{i +pCR), this is taken in 
two steps, xR and x 1/(1 -\-pCR), the second step is a repeat of case (i). 
In case (iii) eJe 1 =pCRl(i -\-pCR), care must be taken with the order 
in which the components XpCR and X 1/(1 -\-pCR) are applied. 

Thus, if pCR(=CRd(dt) is applied first to the step function of ampli- 
tude the result is not obvious, as an infinite rate is encountered, but if 
1/(1 -\-pCR) is applied first, using case (i) this difficulty is removed and 
the differentiation can be performed by inspection. On the other 
hand, with a step function of velocity it pays to apply XpCR first as 
indicated, the differential of the input being obvious. Similar remarks 
apply to case (iv), which is different only by a constant LjCR. Case (v) 
is again similar, and case (vi) is a repeat of case (i). 

The simple case (vii) is a very important one ; it is the basis of nearly 
all linear time-bases, the integral (i//>) is obvious in both cases. Case 
(viii) is occasionally used in a similar capacity for generating linear 
current wave-forms for magnetic deflection of a cathode ray tube, but 
it suffers from the disadvantage that L cannot be had without series i?, 
see case (ix), which is similar to case (vi) with a scale change. 

Case (x) illustrates the application of the method to a more complex 
circuit. Here the transfer function 



*i 



*i 



x 



i +pCR t 



R x +R 2 i+ pC{R 1 R 2 jR 1 + * 2 ) 

x (i +pCR& 



x 



is split as follows 

_ *i 
K R, + R 2 * " i + pCiR.RJR, + Rj 

and the final output is the sum of two separately determined com- 
ponents. These methods, which avoid the solution of differential 



334 



TIME BASES 






equations, are applicable whenever the operator can be split up into sums 
or products involving differentiation (p), integration {ijp) and not more 
than one factor of the form i /(i +pT). If two such factors are present the 
resultant differential equation must be solved for the particular case, 
as there is then no single easily-remembered solution. The same 
remark applies to circuits containing L, R and C, for here again second- 
order expressions in p appear in the denominator. In spite of these 
restrictions the method has wide application, particularly since, by 
superposition of components, the response to many other wave-shapes 
can be determined. 



t (input) 



^ 



e (output) 



R i 



*•**& 



1 



1 



A * \ + = 

/ jf \RT\ 

—f^C e (output) 

Fig. 226 



V 




Derivation of Compensating Networks 

Another use of the method lies in the derivation of networks to com- 
pensate for some unwanted but inevitable feature of another network. 
A case in point is where it is required to generate a linear rise of current 
m an inductor for a magnetically deflected cathode ray tube, in response 
to a step function of amplitude. Here case (viii) might be used if it 
could exist, giving the operator i/(Lp). In fact, case (ix) is all that can 
be had and its operator can be re-expressed as 

1 1 

Lp i+rj{Lpj 
Evidently to obtain the desired result we should first pass the step 
function through a network defined by the operator (1 +if(Tp)) t 
where T= Ljr. Fig. 226 shows such a network, the step function being 
one of current. The output potential is as shown, and when applied 
to the coil, through a cathode follower, to avoid loading the network 
the current in the coil is 



/ X 



R 

Lp 



V + RCp)\\ 



1 



+ r/Lp 



/T 



RI 

"Lp 



I_ 

rCp 



APPENDIX IV 



335 



This represents a linear rise of current at the rate l(rC), amperes per 
second, which is the required result. 

Use of these Methods in Circuits containing Non-linear Elements 

It is important to remember that the above results and methods apply 
only in linear circuits : this does not mean that they must not be applied 
to circuits containing non-linear elements such as diodes, it means that 
they must be applied to such circuits only during those periods in 
which the circuit operates in a linear manner. There may be several 
such periods in each complete cycle, each period being governed by 
different temporarily linear parameters. 

Table XVII has been compiled to show the result which would be 
obtained with perfect differentiating and integrating networks when the 
input takes the form shown in Figs. 216, 218 or 219. 

Table XVII 



Differentiation 



♦Vertical lines above and below 
the base line alternatively . . 

♦Vertical lines on one side only 
of the base line 



Square wave on one side only 
of the base line 



Square 



INPUT 



SQUARE 



SAW-TOOTH 

with zero fly- 
back time 

SAW-TOOTH 

with finite fly- 
back time 

TRIANGULAR 



Integration 



Triangular 



Parabolic 



Parabolic 



Pairs of parabolic 
half-cycles 



* In theory, these lines extend to infinity. 

The results shown in Table XVII can only be obtained if the differ- 
entiation or integration is perfect but, nevertheless, they give an 
indication of the tendencies of the circuits described in this appendix. 

Phase Shift and Attenuation in a Simple Differentiating Network 
having a Sinusoidal Input 

For a sinusoidal input, the phase-shift is 



B = tan -1 



.'/■' 






336 



TIME BASES 



where 6 is the phase lead in degrees, « is 2*/, / the frequency, and 
T=CR or LjR seconds. 

It will thus be seen that the phase-shift B is the angle whose tangent 
is numerically equal to -L and B is the angular relationship between 

the impedance of the shunt arm of the network and the total complex 
impedance. 

Note that the phase-shift of the circuit composed of resistance and 
capacitance in series, is such as to make the output potential lead 




10" J5° 20" 25" 30° 35" 40" 45" 50" 55" 60° 65° 70° 75° 80° 85° 
Angle btf which e leads with respect toE 

Fig. 227. Curve showing the phase-shift in a simple differentiating 
network for a sinusoidal input 

with respect to the input, since the potential across the resistance is in 
phase with the current through the condenser, and this current leads 
with respect to the potential across the condenser. 

In the equivalent circuit consisting of resistance and inductance in 
series, the current lags behind the potential across the inductance 
so that the output potential leads with respect to that across the 
resistance. 

A curve showing the phase-shift 6 plotted against «>T appears in 
Fig. 227, from which the resultant angle of lead may be obtained. 



' 






APPENDIX IV 337 

The ratio of the output potential e to that of the input, for a sinusoidal 
input, is given by 

I - cos 6, 

where e is the output potential and is the phase-shift. This ratio has 
been plotted in Fig. 228. 

The above equation may be written in Heaviside operational form as 

E 1 +pT 

and this may be used when the input wave-form is a unit, or step 
function. 



wT 

10 
$ 

8 
7 
6 
5 

4 

3 



7 
0-9 
0-8 
07 
06 
OS 

04 
OS 

0-2 

































— 










— 






















































































I 




























1 — 

ft 
























— II f o-w-t-o 
S R% e E X| e 

i— J— 

T-CRorjf e=tan' % -^f 

e m pCR _ 

E -coseor 1+pCK i 
































































or 


P U /R 
1+pty 


-1 





0-1 0-2 0-3 0-4 0-5 0-6 01 0-8 0-9 7-0 

%. 

Fig. 228. Curve showing the attenuation of a simple differentiating 
network for a sinusoidal input 

Phase Shift and Attenuation in a Simple Integrating Network 
having a Sinusoidal Input. 
For a sinusoidal input, the phase-change istf = tan =1 co7\ 
It will thus be seen that the phase-shift B is the angle whose tangent 
is numerically equal touT. is again the angle between the output 
impedance and the total impedance. 
24 



338 



TIME BASES 






Note that the phase-shift is such as to make the output potential 
lag with respect to the input since the potential across a condenser 
lags with respect to the current flowing into it. 

Similarly, the current through an inductance lags with respect to 
the potential across it and, since this current flows through the resis- 
tance, the output potential lags with respect to that across the inductance 
or across the complete circuit. 

A curve showing the phase-shift plotted against <*T appears in 
Fig. 229, from which the resultant angle of lag may be obtained. 



WO 


1 
































1-0 
































7 


4-0 


































W 


































H 


































10 


































fM 




















O-JWWW^-p 


— 

e ■ 
-0 








0-4 




















1 

- T = CR 
8 *>tan,~ T u>'i 








<M 


























H 


























0-1 


/. 




















■ 













Afiq/e btf which e fags with respect toE 

Fig. 329. Curve showing the phase-shift in a simple integrating network for a 

sinusoidal input 



The ratio of the output potential to that of the input, for a sinusoidal 
input potential, is given by 

e 

— = cos 0, 

E 

where e is the output potential and 6 is the phase-shift in degrees. 
This ratio has also been plotted against t*T in Fig. 230. 

In the case of a unit function input wave-form, the above equation 
takes the form 

e _ 1 

E ~ T+~pf' 






(JT 






APPENDIX 


IV 








WO 






















hi) 






















4-0 




















±0 






















2*0 




















t-o 




< 


R 

: — vwwv^-i — < 

% cm 

a 1 — 1 

T=CR , 
$ ^tan'%7 


> 












H 
















0-4 






3 










0-4 
















H 

Of 






1 ^ 


= cos 


6 






• 




a i 



339 



Fig. 230. Curve showing the attenuation of a simple 
integrating network for a sinusoidal potential input 

Whiteley *s Differentiating Circuit 

In employing a passive differentiating circuit, strictly accurate dif- 
ferentiation is obtainable only when the output is of zero amplitude 
and, hence, it is necessary to compromise between accuracy and output. 
In general, accuracy cannot be achieved if any reasonable output is 
required. 

In order partly to overcome this difficulty Whiteley* has developed 
the differentiating circuit shown in Fig. 231. A three-stage resistance- 
capacitance coupled amplifier is employed in order to increase the 
output and/or the accuracy of differentiation. Extra pairs of amplifying 
stages may be added if it is desired still further to increase either the 
output or the accuracy of differentiation. 

It is necessary that the input circuit shall be substantially resistive 
and, therefore, the value of the resistance R x must be sufficient to 
render negligible the effect of any reactance in the source whose 
potential is to be differentiated. 

The amplifier employs an integrating negative feed-back network 
L t R lt and it should be noted that if the feed-back network is in the 
form of a differentiating network, the complete circuit behaves as an 
integrator. The use of an integrating network results in increased feed- 
back and, hence, in reduced amplification at the lower frequencies 
because the impedance between the grid of the first valve and the 

* J. W. Whiteley: British Patent 575250- 



34° 



TIME BASES 



negative rail then becomes greater than that between the anode of the 
final valve and the grid of the first valve. It is essential to ensure that 
the source impedance is negligible by comparison with the value of R x 
so that the input terminals may for this purpose be considered as short- 
circuited. As a result of this action, the changes of slope of the input 
wave-form are amplified to a greater extent than are those portions of 
the wave-form whose slope is constant. Reference to Chapter IX will 
show that Miller-capacitance time bases, which are, of course, integrating 
circuits, have differentiating negative feed-back circuits usually con- 
sisting of a condenser and a resistance. 



°HT.+ 




u 

Fig. 231. Whiteley's differentiating circuit 

The output potential is thus integrated by the circuit LJl^ and is fed 
back to the input circuit with an amplitude which is approximately 
equal to the potential applied to the input terminals but of opposite 
phase. Hence, the total input to the valve V x is very small. Triode or 
pentode valves may be employed. 

This circuit is, of course, based on the fact that a valve having a 
resistive anode load, together with an inductance L, connected between 
the anode and the grid, behaves as though an inductance AL were 
connected, in series with a resistance i/g, between the grid and the 
negative supply rail, where A is the stage gain and g is the mutual 
conductance of the valve. The circuit is thus a type of "Miller" circuit 
since it makes use of the Miller effect. 

The Miller -capacitance Integrator 

Miller* found that the stray capacitance between the anode and the 
grid of a valve having a resistive anode load results in the appearance 
of an increased effective capacitance between the grid and the cathode. 

• J. M. Miller: "Dependence of the Input Impedance of a Three-electrode Vacuum 
Tube upon the Load in the Plate Circuit," Set. Papers of the Bureau of Standards, 
1919, 15, p. 367; F. M. Colebrook: "A Generalized Analysis of the Triode Valve 
Equivalent Network," jfourn. I.E.E., 1939, 67, p. 157; C. J. Mitchell: "Miller Effect 
Simplified," Electronic Engineering, 1944, 17, p. 19. 












APPENDIX IV 



34 1 



This effective capacitance has a value C EFF =-(A+i)C AG in series with 
a resistance equal to i}g where C AG is the anode-grid capacitance and 
A is the stage gain. This effective capacitance appears in parallel with the 
already existing stray grid-cathode capacitance C 5 , as shown in Fig. 232. 

uR 

Obviously, if the valve is a triode, A = l , where R a is the 

Ra ~T -»M 

internal resistance of the triode but, where R^ R a > as is usual for a 
pentode, A = gR v 

~otfT.+ 




Czff 



-oH.Z- 



Fig, 232. The effective input capacitance of a Miller integrator 

Blumlein* has developed an integrator circuit, based upon Miller's 
discovery, in which a large degree of negative feed-back is introduced 
by the deliberate application of added capacitance between the anode 
and grid of a pentode valve. The circuit arrangement is shown in 



oHX+ 



o Output 




Input 



oH.T- 



FlG. 233. Blumlein's Miller-capacitance integrator 

Fig. 233 and the appropriate wave-forms appear in Fig. 234. The 
equivalent circuit of a Miller integrator is shown in Fig. 236. The 
cathode of this integrator normally has a bias resistor of such a value 
that the output potential e 2 is, say, 250 volts or more when the input 

*A. D. Blumlein: British Patent 580527; F. C. Williams and N. F ; Moody: 
"Ranging Circuits, Linear Time Bases and Associated Circuits," Journ. I.E.E., 1946, 
93, Part II I A, p. 1189; F. C. Williams: "Introduction to Circuit Techniques for 
Radiolocation," jfourn. I.E.E., 1946, 93, Part IIIA, p. 289; C. L. Faudell: British 
Patent 51 1847; L. R. J. Johnson; British Patent 598304. 



342 



TIME BASES 



potential is zero. The application of a Heaviside Unit Function* (see 
Fig. 234 (a) ) at the input terminals causes a current EjR 1 to flow in R v 
The stray grid capacitance C s prevents the grid potential from rising 
immediately, but it does tend to rise and causes that of the anode e z to 
fall. When the anode potential falls, it drives the grid negative until an 
equilibrium condition arises in which the current EjR 1 flows into the 
condenser C x and the grid potential remains sensibly, but not quite, 
constant. 




1 (a.) 

Input potential 
to grid circuit. 



Bt 

?R (b) 

Output potential 

*2 



FlG. 234. Wave-forms pertaining to the circuit of Fig. 233 



The potential at the grid rises at a rate determined by 

C 



(- 



"> 



where A is the stage gain. The rate of rise at time /=o is 

E 

[C X {A + i)| CJ*! 
and the anode potential changes at A times this rate, which is 

AE A E 

[C&A + 1) + CJif l " T+T / c+ C, \ R ' 

Since, especially when a pentode is employed, as is normally the case, 
A is large, this is very nearly equal to y ■= — - R lm The rate of 



(«♦:&) 



rise of anode potential is thus dependent only on the input potential, 
the sum of the anode-grid capacitance, the grid-cathode capacitance 
divided by A + i and the input resistance R v The effective maximum 
potential is AE, so that the linearity is increased by the value of A, 

* The idea of a Unit Function is due to Heaviside and is defined as a function 
whose value is zero at all times before time ( = is constant at unity thereafter. 
O. Heaviside: Electromagnetic Theory, Ernest Benn Ltd., London, 1893. Reprinted 
1925. 






APPENDIX IV 



343 



which may be between, say, 1,000 and 3,000. Therefore, if g=6, 
R x =500,000 and £=100, the effective charging potential is 300,000 
volts and, if the anode potential changes by 200 volts during the linear 
part of the sweep, the error in linearity will be 0-07 per cent (see 
pages 130 and 162). 

The action of the circuit may be considered from a different point 
of view. It should be noted that the feed-back circuit C^ is actually 
a differentiating circuit. The use of a differentiating circuit results in 
increased feed-back and, hence, in reduced amplification at the higher 
frequencies. This results in those portions of the wave-form which are 
of constant slope being amplified to a greater extent than those portions 
in which the slope changes. There is, therefore, a tendency for the 
slope of the output potential to remain constant. 

Obviously, the circuit will function as an integrator whatever the 
wave-form of the input potential. Moreover, the input potential may be 
positive or negative with respect to the negative rail. 

In contrast, a differentiating circuit may be arranged by interchanging 
C x and R lr i.e. by altering the feed-back portion of the circuit to an 
integrator. 

Wave -front Distortion 

In practice, time base wave-forms (and so-called square waves) rarely 
possess sharp corners, this defect being usually brought about by the 
presence of stray capacitance. Two instances of this kind of distortion 
will perhaps be sufficient to show the cause of the effect. The first 
concerns the effect of the elementary integrating circuit, Fig. 216 
(right-hand side), on a linear potential wave-front. This effect is 
obtained by the application of a potential having an ideal linear sloping 
wave-form, as shown in Fig. 113, across the circuit. The applied 
potential wave-front and the calculated form of the resulting output 
potential are shown in Fig. 235 for various ratios of the time constant 
of the integrating network, denoted by CR, to the duration of the 
applied wave-front T. It will be observed that the output potential 
continues to rise for some time after the applied potential has reached 
its maximum value; the sharp corners at the beginning and end of the 
applied wave-front are not reproduced and the attainment of any given 
potential level is delayed by a time interval which increases with CR, 
When the time constant of the integrating circuit is small compared 
with the duration of the applied wave-front, e.g. CR=o-2$T, the 
curvature of the output wave-front does not extend very far and there 
is a useful portion, which is substantially linear, running parallel to 
the input although delayed by a small time interval which is approxi- 
mately equal to CR, 



344 



TIME BASES 



Practical examples of the application of a linear wave-front to a 
simple integrating network may be found in the use of a grid stopper 
in an amplifier, owing to the presence of stray capacitance from grid to 
earth, and also, for example, in the circuit of Fig. 239 (Appendix VI) 
which is used for producing square waves from a sinusoidal input. 
The input capacitance of the amplifier valve, multiplied by the value 
of the stopper resistance, gives the time constant CR, and the resulting 
distortion may be estimated from the curves of Fig. 235. 




Frc. 235. 



4 5 
Time (t) 

Response of a simple integrating circuit to an ideal 
linear sloping wave-front 



A second instance of this kind of distortion arises when a linearly 
changing current is applied to a resistance which is connected in parallel 
with a condenser. This situation is found in the anode circuit of a 
pentode which is employed to amplify the output of a time base. 
Although the anode current may change linearly, distortion will be 
introduced by the unavoidable presence of capacitance in parallel with 
the anode load resistance and the resultant rise of potential across the 
anode load will not be linear. Fig. 235 applies equally to this case, CR 
being the value of the anode load resistance multiplied by the shunt 
capacitance. In order to employ Fig. 235 to estimate the resultant wave- 
front distortion, the total change of anode current multiplied by the 
value of the anode load resistance is employed, since this represents the 
change of anode potential and, hence, the amplitude of the input 
wave-front. 






APPENDIX V ^ 
SIMPLE ELECTRICAL COMPUTING NETWORKS 

There are many occasions when it becomes necessary or desirable to 
carry out simple mathematical operations, in an automatic manner, 
upon quantities which already exist in a circuit in the form of steady 
or varying potentials or currents. The requirements exist in some 
automatic control or servo systems, in some radar apparatus, in electric 
computing machines and in electrical circuits arranged to simulate 
mechanical systems. These electrical simulators enable the performance 
of mechanical systems to be studied and the effects of changes of 
parameters can be rapidly determined by adjustment of the component 
values of the circuit instead of by the laborious method of changing 
the dimensions or materials of parts of the mechanical system. Further- 
more, the time scale can be altered to enable mechanical systems 
operating at low speeds to be simulated at higher operating speeds and 
the results may be viewed on a cathode ray oscillograph screen. 

Williams and Ritson* have written an excellent paper in which this 
subject is treated and an abstract from the paper is quoted. Although 
the paper deals primarily with servo simulators, the networks discussed 
may, of course, be used for other purposes. Integrating and differen- 
tiating networks have already been discussed in Appendix IV, but they 
are also included here for the sake of completeness. 

Factors Determining the Choice of Computation Methods in 
Simulators 

In making a simulator for a specified linear control system, the design 
problem is to construct an apparatus whose characteristic equation is 
identical with the specified characteristic equation. Means must there- 
fore be available for carrying out the mathematical operations implied 
in the normal characteristic equations met with in automatic control 
engineering. These operations are: 

(a) Addition and subtraction. 

(b) Integration and differentiation with respect to time. 

(c) Multiplication and division. 

All these functions can be performed mechanically, and, in fact, are 
so performed in the differential analyser, f It is, however, possible to 
perform these operations electrically, provided that the input and 

* F. C. Williams and F. J. U. Ritson: "Electronic Servo Simulators," Journ. I.E.E., 

I9 fV. Bush: "The' Differential Analyser," Journal of the Franklin Institute, 1931, 
212, No. 4. 

345 



34<> 



TIME BASES 



output quantities of the control system are represented by potentials or 
currents in the simulator; it is much more convenient to use electrical 
computation methods as then no precision engineering is involved, and 
only readily available standard components are used. 

Electrical methods can be subdivided into purely electronic methods, 
using no moving parts, and electro-mechanical methods, which may 
include instrument- type servo-mechanisms such as velodynes.* 

An advantage of the purely electronic simulator is the flexibility of 
the time scale it offers; it may be made to operate so fast that the 
response wave-form can be observed on a cathode ray oscillograph, so 
that the effect of changes in the parameters is immediately obvious. 
Alternatively, of course, it may be operated so slowly as to permit 
the use of recorders or manual plotting methods. It should be stressed 
that the time scale of the simulator may have any desired relationship 
with that of the control system it represents ; and its choice is largely 
decided by the method used to record the results. 

Summary of Electrical Computation Methods 

Apart from the fundamental mathematical processes listed in the 
previous section, several more complex operations, such as "smoothing" 
and "phase advancing," are commonly carried out. Since these opera- 
tions are very often performed electrically in the control system 
itself, these same methods can be applied directly in the simulator. 
Table XVIII, which summarizes the methods of performing the 
simpler operations, includes several of these. 

Column I shows purely passive networks, while column II shows 
circuits developed from these which employ feed-back amplifiers and 
have advantages to be discussed later. The operationally-stated transfer 
function is shown for each circuit. Those in column I are well known, 
while those in column II have been derived from those in column I by 
using the fact that the input impedance of feed-back amplifiers of this 
type is an impedance equal to if (A + i), of the feed-back impedance: 
where A is the stage gain, i.e. the change in anode potential per unit 
change in grid potential. The equivalent input circuits for column II are 
then identical with the circuits of the first column, provided A is real. 
The output potential is obtained by multiplying the effective amplifier 
input potential by A. For example, in case (D) the equivalent input circuit 
is as in Fig. 236. The amplifier input potential e g is ej[i +pRC(A + i)], 
and the output potential is AeJ[(i +pRC(A + i)]. The transfer function 
for the circuit is eJe l =Aj[i +pRC(A + i)]. 

The following points about the tabulated circuits are worth noting : 

*F. C. Williams, F. J. U. Ritson and T. Kilburn: "Automatic Strobes and Re- 
currence Frequency Selectors," Jown, I.E.E., 1946, 93, Part IIIA, p. 1275. 









APPENDIX V 



347 



In case (A), column I, the resistance i? 3 may represent the input 
resistance of a circuit performing a subsequent operation, or it may be 
deliberately introduced to reduce interaction between the sources e x 
and e 2 : in this case R 3 must be small and then the attenuation will be 
severe, and some subsequent amplification will be necessary. It is 
therefore often much more convenient to use the corresponding circuit 
in column II, since, by so doing, the effective value of R z is made very 
small, but the precise gain A is not critical, provided it is large, as may 
be seen from the approximate value of the transfer function obtained 
by allowing A to approach infinity. If, instead, the column I circuit 
were used with an amplifier following it, the gain of that amplifier would 
have to be precisely related to the attenuation, a result which could, 



o vwvwv- 



AmpHfter 

of Gain 

-A 



e z 



Fig. 236. Equivalent circuit of Miller integrator 

in fact, be obtained only by the use of internal negative feed-back in 
the amplifier (see, for example, case (B), column II). 

In case (B) passive networks only provide division by constant ratio, 
whereas the column II method can also provide multiplication (amplifi- 
cation) in a fixed ratio. 

In case (C), column I, the closeness of the approximation to true 
differentiation increases in given circumstances, as the value of R is 
reduced. Unfortunately the output potential available decreases as R is 
reduced, and once again some amplification is necessary. The circuit 
of column II is therefore often preferable for similar reasons to those 
obtaining in case (A). 

In (D) the column I circuit is very useful for its true function of 
smoothing undesired ripple, and for representing the exponential delay 
characteristic of servo field windings. When this same circuit is used 
as an approximation to integration for the purposes of integral control, 
or to represent, for example, the integration of torque into velocity 
performed in practice by the rotor inertia of a servo motor, the output 
potential will be very small, since the approximation to integration is 
good only when the components are such that the output potential is 
negligible compared with the input potential. Once again, in such cases 
the circuit of column II is much to be preferred : this is the well-known 
"Miller integrator." 



S48 



Mathematical 

Operation 



Addition 

and 

Subtraction 

Case A 



Division 

by 

Fixed Ratio 

Case B 



Differentiation 
with respect 
to Time 

Case C 



Integration 

with respect 

to Time 

Case D 



Phase -Advance 
Case E 



Error 

+ Integral 

of E rror 

Case F 



TIME BASES 



Table 



Column I 
Circuits consisting of Passive Networks 



R, 

O VNAA— 



cww- 



e 2 



SK. 



-o 
63 




L 





Neglecting source 
impedance ofe, and e 2 

= RaiSfKi+esXi) 

3 R,R 2 +R 2 R 3 +R,R 3 

if R 3 « R, and R 2 



e, R, +j?2 



e?_ pCR 
e,~i + pCR 



e, i + pCR; 






fz I + pCR 2 

e, J + PC(R, + R ? ) 






APPENDIX V 



349 



XV1I1 



Column IT 
Circuits using Feedback Amplifiers 



J? 

O-AAA, 1 



ft 



e 



1 — W 1 

1-I--0 o » O 



i- Amplifier gain = —A 




Amplifier gain = —A 



r 1 

t 



I — i/WV 

I I -O O-* 

I 



Amp// Tier ga/n = —A 



-fh 



&AV- M-o 



1 



Amplifier gain =—A 



R, 




e, r 







Amplifier gain = —A 



R, C 
t-W — |H 



R 

I 



I 



4 



Amplifier gain — —A 






AR 3 



e, * 2 +iZ,</+-A) 



*W 






7 + pC 



r+A 



B\ t + pCRiit+A) 



%-A- 



R, 



t+pCR 



e,' Rr+RO+A)^^ 



Rj±R(1+A) 



e 2= ; + jpCR, 

ej 7+pc{r } +R(A+i)} 



If A »•?, We/? 



IPA»7, then 






& 

*; 



ffAy>7, then 



^^-pCR 



IfA» T , then 



e, 



'pCR, 



//M » 7, t/?e/7 



e ; A 2 



IfA» 1 , then 
e, />CK 



35° 



TIME BASES 






In case (E) the column I network is preferred because the transfer 
function of the column II arrangement depends on amplifier gain for 
low values of A, and the low-frequency attenuation of the column I 
arrangement is rarely more than 10. Alternatively, the arrangement of 
Fig. 237 may be used, in which, if A !>i the transfer function is sensibly 
independent of A : the advantage of this arrangement is that the two 
time-constants can be adjusted separately. 

In case (F) the column II arrangement is preferred for the same 
reasons as in case (D). 

-O H. T. + (Screen) 

O+300V 




ez i? 2 (1 + pCM .. 

« « % ■ (T+ pcjQ lf sta s e & m A > l 

Fig. 237. Phase-advance circuit, with smoothing 

AH the approximate transfer functions listed for column II can 
be derived by inspection, using the method in which the 
amplifier input terminal is regarded as being held at earth potential by 
the operation of the amplifier. The output potential can then be calcu- 
lated by calculating the current which would flow in the input network 
and assuming that the output potential is due to this current flowing 
past the "virtual earth" through the feed-back impedance. 

Various electronic methods of multiplying together two potentials e 1 
and e 2 have been devised, but these tend to be too complicated or 
inaccurate to be used in simulators. 






APPENDIX VI 
THE GENERATION OF SQUARE WAVES AND PULSES 

There are many methods of producing square waves and they may 
be generated directly or obtained by the modification of wave-forms 
of other shapes. 

A convenient method of generating a square wave is to obtain 
it from the free anode of a pentode multivibrator in which oscilla- 
tion is obtained by means of feed-back from the screens to the grids 
(see page 31) or from a flip-flop (see page 64). Alternatively, for 
low frequencies such as 50 cycles, a sinusoid may be squared by 
applying it to the grid of a valve whose grid base is small compared 



OH.T+ 




OOut 



Sinusoidal 
Input 



OH.T- 



Fig. 238. Method of generating a square wave 

with the amplitude of the sinusoid.* The resulting wave-form is, of 

course, a very poor simulation of a rectangle, but for many purposes 

a chopped sine wave is sufficient. Fig. 238 shows a suitable circuit for 

squaring a sinusoid. The diode V x is employed to prevent the flow of 

grid current so that an input of 100 volts or more may be employed. 

The value of the resistance R x should be sufficiently large to limit the 

peak current to a value within the rating of the diode, say, between 

2 and 10 milliamperes. Since current flows only during the alternate 

E 2 
half-cycles, the required power rating of J^is £ watts, where E is the 

1 

R.M.S. value of the potential across the secondary winding of the 

transformer. The value of R t should not be made greater than 
necessary since, otherwise, the grid circuit time constant may become 
sufficiently large to have an appreciable effect on the wave-form. 

As a further alternative, but somewhat less satisfactory, method, the 
diode V 1 may be omitted and the resistance R x connected in series 

* A device of this type is commonly termed a " limiter, " " clipper" or "chopper. " 

35i 



352 



TIME BASES 






between the transformer and the grid of the valve F 2 , as shown in 
Fig. 239. The value of R t is, in this case, best determined by experi- 
ment. Too low a value of resistance will result in an appreciable 
variation of grid potential during the period when the grid is driven 
into grid current. Too high a value will reduce the steepness of the 
wave-front. A small condenser of about 10 to 20 micromicrofarads 
connected in shunt with R x may be found helpful in reaching a satis- 
factory compromise, R t being adjusted to a value between, say, 
100,000 ohms and 1 megohm. 

When it is required to produce a square wave-form, it is sometimes 
also necessary to predetermine the duration of the wave-front. If the 
input potential is sinusoidal it is easy to calculate the time required for 
the input potential to change by an amount equal to the grid base of 
the triode valve in Figs. 238 or 239, assuming that it is only during this 
time that the anode current will change. 

OHX+ 




$wusoidaf 
Input 



OH.I- 



FiG. 239, Alternative method of generating a square wave 

The time interval T 2 required for the potential to change from zero 
volts to —E 2 volts, or vice versa, is given approximately by the following 
equation,* provided that E % is small compared with £ L : 



?* = 



E 2 T, 



or 



—2 = - 1 volts per second, 
J 2 ■* 1 






where T 1 is the duration of one cycle and E 1 is the peak voltage of the 
sinusoid. Thus, if for E 2 we insert the grid base of the valve F 2 , in 
volts, the time required for the potential to sweep across the grid base 
and, hence, the approximate time required for the anode current to 
rise to a maximum may be calculated, 

* A more precise equation for the value of T 2 is 

r a = Tisin" 1 ^. 
" 360 ^1 

This should be employed instead of the approximate equation given above when a 
sinusoid having a peak potential which is less than 20 or 30 times the grid base of 
the valve is employed. Only when the amplitude of the sinusoid is very large compared 
with the grid base can the rate of change of potential be considered to be constant 
during the period T 2 . 



APPENDIX VI 






353 



If the gain of the amplifier is denoted by A, the rate of change of 
potential across R % is 

, ! — volts per second 
■* 1 
during the wave-front portion of the resulting square-wave, provided 
that the stray capacitances are of negligible value. 

The presence of stray capacitance between the anode and the cathode 
of F 2 reduces the rate of change of potential, rounds off the otherwise 
sharp corners of the wave-form and introduces a time delay. These 
effects become serious when designing an amplifier to handle a steep- 
fronted wave-form if the duration of the rising portion is of the same 
order as the time constant of the anode circuit. It will be obvious 
from the above equations that, for a given valve grid base, the duration 
T 2 of the wave-front is inversely proportional to the peak voltage E 1 
of the sinusoid. Thus, if the conditions are such that the duration of 
the wave-front is still too great after the squaring operation, a second 
stage may be added to reduce the duration of the wave-front still 
further or, alternatively, the value of E x may be increased. Further 
amplifying stages may, however, fail to steepen the wave-front beyond 
a certain degree unless the anode circuit time constants are progressively 
reduced. In any case, each amplifying stage should be so biased that 
the rounded corners of the wave-form produced by the preceding stage 
are not transmitted. 

It will thus be evident that the circuit of Fig. 238 is very similar to 
that of Fig. 239 so far as distortion of the wave-form at the anode is 
concerned. "Where a rapid rise of the wave-front, as seen at the anode, 
is required, it is advisable to employ the circuit of Fig. 238, while if a 
rapid trailing edge is of greater importance, then the circuit of Fig. 239 
is to be preferred. 

These two circuits generate square waves having a mark-space ratio 
of approximately unity. In practice, the anode potential is at a maximum 
value for a period which is slightly greater than that for which it is 
a minimum. 

When it is desired to generate square waves at higher repetition 
frequencies and with shorter duration of the discontinuity, other circuit 
arrangements are advisable (see, for example, Chapters V and IX). 

Fig. 240 shows a method of limiting the potential excursion in each 
direction by means of a pair of inversely connected diodes which are 
biased to appropriate potentials. The ideally flat top and bottom of the 
square wave is not fully realized in practice with a sinusoidal input and, 
in fact, a segment of a sinusoid is superimposed. This segment has a 

maximum amplitude of E x 



R, 



Ri~\- R& 



25 



354 TIME BASES 

where E x is E— v, 

E is the peak potential of the input sinusoid, 

v is the half-amplitude of the ideal square wave, i.e. E—E B , 

R D the forward (conducting) resistance of the diode, and 

R x the series load resistance. 

In order to keep the top and bottom of the square wave as flat as 
possible R x should be very large compared with R D . The peak-to-peak 
value of the resultant square wave is approximately E B1 + E B2 . 



o— ww- 



Input 



X 



u 



-o Output 



-o Earth 



t 



-OS+ 



-O Output 



-OS, 



Ri --- 
o — ww—» 



■B 2 



°+E3 2 

Fig. 240. Inverted diode limiter 



Input t 

O 



n 



-OJS- 



Fig. 241. Pentode limiter 



Another valuable circuit arrangement is shown in Fig. 241. In this 
case a pentode is employed, and its anode is arranged to "bottom," 
i.e. by the use of a high value for the anode load resistance, the anode 
potential is reduced to the minimum possible value E B while the grid 
potential is still negative, so that a reduction in the negative grid potential 
has no further effect on the anode potential. In this way, the segment 




Fig. 242. Output wave-form obtained with the circuit of Fig. 241 

of a sinusoid which, when using diode limiters, appears on the flat 
portion of the square wave is inhibited for the positive excursion of the 
grid while cut-off of the anode current inhibits the effect for the 
negative excursion (see Fig. 242). That this is so, may be seen by an 
examination of the E A fI A characteristic of a pentode valve which is 
shown in Fig. 243. It can be seen that, provided the load-line for the 
anode load resistance has a sufficiently low slope, the load-line will cut 



APPENDIX VI 



355 



a part of the characteristic below the knee of the curve at a point 
where a reduction of the grid bias is unable to reduce further the 
potential appearing at the anode. 



OV.(E 9t ) 




Anode Volts 
Fig. 243. Anode potential-anode current characteristic of a pentode valve 

Cathode- coupled Transitron Square- wave Generator 

A cathode-coupled transitron flip-flop is described on page 70, and 
by simple circuit modifications it can be caused to function as a square- 
wave generator. Since the basic principle of the circuits is the same, 
reference should be made to the operation of the flip-flop form of 
the circuit. 

The square-wave generator circuit is shown in Fig. 244, from which 
it will be seen that the output can be taken from the anode, screen grid 

g+200 V 



Outo 
Outo 




•oOut 
°+28V 



o+Bias 



o-ZOOV 
Fig. 244. The cathode-coupled transitron square-wave generator 



or cathode. If the output is taken from the cathode, a low-impedance 
source, whose phase is the same as that of the anode, is provided. The 
anode and screen outputs cannot be used as a push-pull source, since 
the wave-forms differ somewhat. 



35 6 TIME BASES 

The static curve for this circuit is shown on page 70. Normally, the 
bias potential is adjusted to the point on the negative slope portion of 
the curve which provides a square wave having a mark-space ratio of 
unity. This point lies near the middle of the curve. Alteration of the 
bias towards the points A and B will alter the mark-space ratio. 

Appropriate circuit values are as follows: 



Rx 


20,000 ohms 


c, 


as required 


R?, 


10,000 ohms 


Vy 


VR116. 


R* 


100,000 ohms (max.) 






*4 


100,000 ohms 







In general, a square-wave oscillation requires the existence of two 
time constants and this is particularly obvious when the mark-space 
ratio is not unity. In this circuit, the second time constant is obtained 
by a modification of the value of the physical time constant. The 
physical time constant has an approximate value of (R x -\-R t >)C v which 
applies when the anode current is cut off. When the anode current is 
flowing (grid volts less than +34 volts), the time constant becomes 
A(R i -rR 2 )C 1) where A is the gain of the stage. 

This circuit is due to Fleming- Williams.* 

Goldmuntz and Kraussf have developed a number of cathode- 
coupled two-valve limiter circuits. 

White's Multivibrator Pulse Generator 

White J has described a multivibrator pulse generator in which the 
grid leaks of the two valves are connected to a high value of positive 
potential. 

The two valves are pentodes and the anodes are coupled to the 
opposite grids by condensers and resistances, the time constant of 
these couplings being of a small value. The anode load resistances 
require to be of a high value to cause the anode potential to fall to a very 
low value and the output is taken from one or both screen grids. 
When the grid of one of the valves becomes positive, the anode potential 
falls and the screen potential increases so that a positive pulse appears 
at the screen grid. A similar negative pulse appears at the other screen 
grid. The screen grids must not have by-pass condensers. 

The Multiar 

The multiar is not a time base but is an amplitude comparator and 

* B. C. Fleming- Williams of A. C. Cossor, Ltd., London. 

+ L. A. Goldmuntz and H. L. Krauss: "The Cathode-coupled Clipper circuit," 
Proc. I.R.E., 1948, 36, p. 1172. 

JE. L. C. White: British Patent 515786. 









APPENDIX VI 



357 



square- wave generator. It was developed by Williams* in 1944 as an 
accurate range-marking device for use in radar.f The circuit is fed 
with a negative-going saw-tooth wave-form and means are provided 
for producing a square wave-form having a wave-front of extremely 
short duration. 

Since the anode current is reduced when the grid potential is made 
slightly negative, the cathode current falls in amplitude and the diode 
V z , whose anode is at a negative potential, has its cathode driven more 

V+330V 

O+2Q0V 



o Output 



o— 



Control ^- 
Potent/a/ p 
o h- 




Earth 



-O-200V 



Fig. 245. The Multiar 



negative than its anode. This drives the grid of the pentode F 3 violently 
negative in a cumulative manner. It will be appreciated that, until V 2 
conducts, the feed-back path through the diode is not complete since 
the latter does not normally conduct. The oscillator executes a single 
oscillation which commences when the amplitude of a controlling saw- 
tooth wave-form has achieved a predetermined value, and this value 
can be precisely adjusted by alteration of the setting of the potentio- 
meter R A . The only uncertainty is the precise value of the potential 
across the diode which corresponds with a given diode current. 

The circuit of the multiar is shown in Fig. 245, from which it will be 
seen that a pentode valve F 3 , having a transformer coupling between the 
cathode and the grid circuits, is so arranged that the resultant positive 

* F. C. Williams : British Patent 583553 ; F. C. Williams and X. F. Moody: " Ranging 
Circuits, Linear Time Base Generators and Associated Circuits," Journ. I.E.E., 1946, 
93, Part II 1A, p. 1 188. 

f Since a number of these circuits can be connected to a single time base without 
mutual interference, a multi-range marker system becomes practicable. The device 
was given the name "Multiar," derived from "Multi-R," meaning multi-range. 



358 



TIME BASES 



feed-back is rendered inoperative by the diode V 2 until the amplitude 
of the input saw-tooth is sufficient to overcome the negative bias applied 
to the anode of the diode by the potentiometer i? 4 . 

During the stable condition, in the absence of the input saw-tooth, 
the grid of the pentode V z draws current via the resistances R 2 and R 3 , 
since the resistance R 2 is connected to a high positive potential. The 
anode of V s is therefore at a low potential. The anode of the diode V 2 
is connected to a negative potential of a value lying between zero volts 



t=0 



t^T 



(a.) Input saw tooth" 




the anode of the 
diode 1$. 



(c) Potential at 
the anode of the 



~v 



-N 



pentode Y3 



(d) Potential at 



the cathode 
1 ' of V A ■ 

Fig. 246. Wave-forms pertaining to the circuit of Fig. 245 

and a potential equal to the maximum amplitude achieved by the saw- 
tooth. As a result, the diode is cut off during the stable regime. 

The saw-tooth wave-form which is applied as a control potential 
should have its steady current component restored by a diode V lt so 
that when its amplitude is zero, the input potential to the multiar is 
also zero. The sense of the restoration is such that the saw-tooth 
potential cannot become positive with respect to earth potential. 

When, at time t=x, in Fig. 246, the amplitude of the saw-tooth 
wave-form is sufficient to cause V 2 to conduct, the feed-back circuit 
is closed and V s is cut off extremely rapidly. The wave-front of the 
resulting square wave at the anode of V z is, therefore, extremely sharp, 
provided the value of i? 5 is not too great and the stray capacitance 
across it is kept to a minimum. The appropriate wave-forms are shown 
in Fig. 246, in which (a) shows the input saw-tooth wave-form, the 
horizontal line representing the negative bias applied to the anode of 
V 2 by means of the potentiometer R 4 , (b) the potential at the anode of 



APPENDIX VI 



159 



the diode V 2t (c) the potential at the anode of the pentode J 7 3 and (d) 
the potential at the cathode of F 3 . 

The initial application of a negative potential to the grid of V 2 as a 
result of the potential of the cathode of the diode V 2 becoming steadily 
more negative and, passing the potential of its anode, decreases the 
potential of the cathode of V 3 . This change of cathode potential is 
stepped up by the transformer 7\ and applied to V. A via the diode F 2 
in such a sense as to make the grid of V 5 still more negative. 

The time constant C^ must, of course, be long enough to prevent 
the potential of the grid of F 3 rising during the cut-off period. 

The fly-back commences at the time t=T, and when it results in the 
potential of the cathode of F 3 passing that of its anode, V 2 is cut off; 
this in turn disconnects the feed-back circuit so that the grid of F 3 is 
driven positive (since it is then once more able to draw grid current 
via R 2 and R 3 ) and the system returns to the stable condition. 

The time constant C X R 3 must allow the whole of the available 
potential to appear at the grid of V 3 if the return of the anode potential 
to the stable condition is to take place rapidly. Care should be taken to 
see that this occurs, especially if the saw-tooth fly-back is of short 
duration. 

The transformer consists of a core of radio-metal having a single 
layer primary winding of 135 turns of 38 S.W.G. copper wire, over 
which is wound a secondary winding of 200 turns of the same gauge 
of wire. The other components may have the following values: 



*M 


270,000 ohms 


c, 


1,000 fitiF. 


Rt> 


10 megohms 


c 2 


o*i txF. 


R* 


470 ohms 


c. 


o-i [iF. 


*4 


250,000 ohms 






£« 


56,000 ohms 


Vi 


EA50 or VR91 


*« 


33,000 ohms 


v 2 


EA50 or VR91 


*7 


1,500 ohms 


V* 


EF50 or 6AC7 



If desired, the saw-tooth input wave-form may commence at a 
positive potential since the grid of F 3 is coupled to the anode of the 
diode V 2 via a condenser. In this case, the bias potential value at the 
anode of V 2 must be correspondingly adjusted to a potential which is 
less negative than before. 

It will be obvious that the time of commencement of the positive- 
going square wave at the anode of V z is controllable by adjustment of 
the potentiometer #4. Sometimes it is required that the square wave- 
form shall be a negative-going one commencing at the time *=o and 
ceasing at time t=x. This can be achieved by applying a negative-going 
square wave between the suppressor grid of F 3 and the negative supply 



360 



TIME BASES 






rail, before the time t=o so that no anode current flows, but grid 
current continues to flow during this part of the stable regime. The 
suppressor-grid potential must then be raised to zero at time t=x 
and kept at this potential until just before the arrival of the next saw- 
tooth wave-form. 

When the negative pulse appears at the cathode of V s at time t=x f 
the effective suppressor grid bias is reduced and the anode current 
flows in F 3 , resulting in its anode current being cut off owing to the 
elimination of feed-back. Thus a positive step having a steep wave- 
front is produced at the instant t—x. 

Two alternative arrangements are possible, (a) in which the primary 
of the transformer is connected in the anode circuit of the valve F 3 
and the output is taken from the screen-grid circuit, and (b) in which 
the transformer is connected in the screen-grid circuit and the output 
taken from the anode. 

Since the time delay from t=o to t=x is determined solely by the 
potential applied to the anode of the diode V 2 via the potentiometer 
j? 4 , it is essential that the —200- volt bias supply shall be thoroughly 
stabilized. Chance* has pointed out that if this supply rail is modulated, 
say, with speech signals, the output will be time modulated, the width 
of the pulses being determined by the instantaneous amplitude of the 
applied signal. 

Many other circuits which produce square-wave output potentials, 
though without that degree of precision which the multiar can provide, 
are described in Chapters III, IV, IX and XII. 

The Inductive Pulse Generator 

In this method of generating a square wave having a mark-space 
ratio which differs widely from unity, an inductance is connected in 
the anode circuit of a pentode valve whose anode current is normally 
cut off. When the anode current is switched on or off, the inductance 
is shock-excited and produces a damped oscillation at a frequency 
determined by the value of the inductance and the total shunt capaci- 
tance. The circuit is shown in Fig. 247, while the relative wave-forms 
are shown in Fig. 248. In the absence of the damping imposed by the 
diode V 2 and by grid current in F 3 , the "ring" produced at the instant 
of switching off the anode current in V x is identical with that which 
occurs upon switching on except that it is reversed in phase. This 
statement may not be quite true if the oscillation does not die out 
during the "mark" period, but does do so during the "space" period, 
or vice versa. 

* B. Chance: "Some Precision Circuit Techniques used in Wave-form Generation 
and Time Measurement," Rev. Set. Insts., 1946, 17, p. 396. 



APPENDIX VI 



361 






When the current is switched on by a positive pulse at the grid of 
V lt the first surge of the anode potential is negative, and this is applied 
to the grid of V s which is normally biased to zero volts. The effect of 
the diode is to absorb the positive swings of the anode potential and 



-o+HJ. 




oOut 



-H 



Control 
Potential 



o-H.T. 



Fig. 247, The inductive pulse generator 



Grid of % 




Anode of V t 
{Undamped) 



Anode ofty 
(Damped) 



Anode ofV$ 






Fig. 248. Wave-forms pertaining to the circuit of Fig. 247 



the oscillation becomes effectively damped out from the instant when 
the anode potential tries to go positive with respect to the H.T. supply 
rail. The whole of the oscillation is suppressed by the diode when the 
grid of V x is cut off, but the first half-cycle survives when V x grid is 
driven in a positive direction. Grid current in V % may assist the diode 
in dissipating the energy stored in the inductance, but this is undesirable 



362 



TIME BASES 



because grid current will alter the charge on C 2 and so produce an 
overshoot on the trailing edge of the output pulse unless C 2 is very 
large. Grid current can be reduced by increasing the value of the 
grid stopper. 

The amplitude of the undamped oscillation should be much greater 
than the grid base of F 3 , so that the grid may be cut off for a large 
fraction of the period and so that the resulting pulse at the anode of V 3 
may be flat-topped and steep-sided. The amplitude of the oscillation 
appearing at the grid of K 3 is proportional to the change of current in 

V 1 and to the value of / . It is not advisable to control the pulse 

width by adding capacitance across the inductance since this reduces 
the amplitude of the oscillation at the anode of V 1 and, therefore, the 
sides of the output pulse become less steep. 

The duration of the half-cycle pulse is equal to - SLC, where C 
denotes the total capacitance consisting of the sum of the anode-cathode 
capacitance of V lt the diode capacitance, the self capacitance of the 
coil and the grid capacitance of V s , 

Since the sides of the output pulse are derived from a sinusoid, their 
slope tends to decrease when the oscillation period is increased and, 
for this reason, the method is not recommended when it is desired to 
produce pulses of long duration and relatively instantaneous rise and 
fall. In such cases, a flip-flop is to be preferred, but the inductive 
method is very useful for producing pulses of short duration, say, from 
1 to 10 microseconds. This circuit is due to Whiteley.* 

Blocking Oscillator Pulse Generator 

Whitef has described a blocking oscillator form of pulse generator, 
the circuit of which is shown in Fig. 249. This pulse generator func- 
tions in much the same way as that of Fig. 64, except that a pentode is 
employed and, when the grid is driven positive during the fly-back 
period, the screen current increases and a negative pulse appears at the 
output terminal. 

The valve V Y should have a high value of anode load and should be 

capable of passing a high value of anode current when the anode 

potential is at its minimum value. This can be achieved by applying a 

positive potential to the grid by virtue of the grid resistance R 2 being 

connected to a high positive potential. The high value of anode load 

impedance results in the possibility of reducing the value of the 

resistance R 2 and hence of the time constant C x R 2i which determines 

the repetition period. 

* J. W. Whiteley: A. C. Cossor Ltd., London. 
f E. L. C. White: British Patent 515786. 






APPENDIX VI 



363 



This pulse generator may also be used as a time base by connecting 
another condenser between the grid winding of the transformer T x and 
the cathode of the valve, and by using the saw-tooth which appears 
across the condenser. The terminal marked "Out" in Fig. 249 may 
then be connected to the negative supply rail in order to hold steady 
the potential applied to the screen grid of V v 

The insertion of a large inductance between the resistance R 2 and 
the slider of the potentiometer R z will increase the linearity of the saw- 
tooth potential, and this may prove useful when the circuit is employed 
as a time base. 



o//.Z> 




Out 



OH.T- 

FlG. 249. White's blocking oscillator pulse generator 

The Use of a Cathode Ray Tube for Generating Wave-forms 

If the fluorescent screen of a cathode ray tube is replaced by radial 
metal strips connected together at the centre, the tube may be used for 
the generation of square waves. The cathode ray beam may be caused 
to rotate, as described on pages iTand 101, and a resistance may be 
connected between the centre of the strips and some other point at a 
convenient potential. The potential drop which appears across this 
resistance when the rotating beam falls upon successive strips may be 
amplified and applied as required. 

A method which enables any wave-form to be reproduced from a 
template cut on a piece of opaque material has been described by 
MacKay.* Referring to Fig. 250, a cathode ray tube has a linear time 
base potential applied to the X deflector plates and the light from the 
screen is collected by a lens L and directed on to a photo-cell P. The 
output of the photo-cell is applied to an amplifier whose output is 
connected to the Y deflector plates in such a sense that increased light 
drives the spot behind the opaque mask. 

If the amplifier gain is sufficient, the photo-cell screened from all 
light other than that coming through the lens, and only the light from 

* D. M. MacKay: "A High-speed Electronic Function Generator," Nature, 1947. 
159, p. 406. 



364 



TIME BASES 






APPENDIX VI 



3 6 : 



the cathode ray tube is allowed to enter the lens, the centre of the spot 
is constrained to follow the edge of the mask. It will be obvious that 
a tube with a very short afterglow time should be used since any after- 
glow will affect the position of the spot. It will sometimes be found 
that small variations appear superimposed on the wave-form as dictated 
by the shape of the mask. This is a noise effect introduced by the 
granular nature of the screen material since, in the limit, the light 
emitted by different elemental areas of the screen material will differ 







Fig. 250. MacKay's template wave-form generator 



slightly even though the H.T. potential and the beam current through 
the tube remain constant. The required wave-form is taken from the 
output terminals of the amplifier. This scheme is, in general, more 
simple than that referred to on page in. 

The Use of a Line for the Formation of Brief Pulses 

If it is desired to produce a pulse having a duration of the 
order of a few microseconds, a Unit Function,* as shown in Fig. 25 1 , 
may be produced at the input terminals of a short-circuited low-loss 
transmission line, but, in practice, this is most inconvenient since the 
length of line required is unduly long. For most purposes it is satis- 
factory to employ an artificial linef consisting of series inductances and 

, j shunt capacitances. In these circum- 
stances, the unit function will travel 
down the line and be reflected back 
again in opposing sense, from the 
short-circuited end and so return to 
the input. The time required for the 
signal to return to the input end of the line is dependent upon the 
electrical length of the line. When the signal is reflected, it is also 

* The idea of a unit function is due to Heaviside and is defined as a function whose 
value is zero at all times before t = o, and whose value is constant at unity thereafter. 
O. Heaviside: Electro-magnetic Theory, Ernest Benn, Ltd., London, 1803. Reprinted 

tA. D. Blumlein: British Patent 528310: A. D. Blumlein, H. E. Kallmann and 
W. S. Percival: British Patent 517516. 



t~0 

Time — *- 

Fig. 251. Unit function 






reversed in polarity, so that the resultant wave-form at the input 
end of the line is a square wave whose duration is twice the time 
taken for a signal to travel from the input end of the line to the 
short-circuited end. A further line may also be employed to provide 
another square wave at a time which is delayed with respect to the first. 
It should be noted that if the far end of the line is open-circuited the 
reflected signal does not suffer a reversal of phase. As a result, the 
signal which appears at the input end of the line consists of two unit 
functions of the same polarity, of which one is delayed with respect to 
the other. This delay is equal to twice the electrical length of the line, 



■o/f.T.-h 




Input 



oOut 1 
*-oQut.2 



oH.T.- 



Fig. 252. The use of a line for the formation of square waves 

so that the resultant signal has the appearance of two steps of a scalari- 
form wave. 

If the far end of the line is terminated in its characteristic impedance, 
no reflection occurs, since the whole of the energy in the line is absorbed 
in the terminating load. 

A circuit for producing the two square-wave pulses, of which one is 
delayed with respect to the other, is shown in Fig. 252. The part which 
produces the initial square-wave pulse is marked A and that which 
produces the delayed pulse is marked B. 

In order to design the line, the capacitances C are preferably made 
equal to the input capacitance of the valve V % plus any additional stray 
capacitance. In this way, the capacitances of each section may be made 
equal by omitting the condenser at each point to which a valve is 
connected. If valves having different input capacitances are employed, 
the value of capacitance which is the greater should be taken for the 
capacitance per section of the line and, where the valve having the 
smaller capacitance is connected, a padding condenser may be employed 
to make up the total. 

Thus, in the case of Fig. 252, the value of C will be determined by 



3 66 



TIME BASES 



the capacitance across the grid of the valve V 2 because this valve has, in 
addition to the inter-electrode capacitances, the stray capacitances of 
the condenser C 3 to earth and that of the anode of V x to earth. For 
this reason it may be necessary to add a padding condenser across the 
grid of F 3 to earth to make the total up to the value C. 

The following procedure should be adopted for the design of the 
line: R is usually of the order of 3,000 to 5,000 ohms. 



Then 



L 
t 

N„ = 



= Ci? 2 
= VLC 

w 

zt 
D 



where 



t is the time delay per section, 

N A the number of sections in part A of the line (Fig. 252), 
N B the number of sections in part B of the line, 
W the required width (of the square wave) in microseconds, and 
D the required delay (between the first and second square 
waves) in microseconds. 
It will be obvious that the number of sections in the part of the line 
marked A will determine the duration of the square-wave since it is 
at the short-circuited end of the line that the signal is reflected. The 
number of sections in the part marked B determines the delay between 
the square-wave appearing at the anode of V 2 and that at the anode 
of F 3 . The part marked C is added in order to provide proper termina- 
tion of the line. In theory, it should be possible to dispense with 
section C and to connect the terminating resistance at the grid of the 
valve V 3 provided that the input capacitance of the grid can be made 
equal to \C> but, in practice, this is generally impracticable and, in 
the interests of a better resultant wave-form, it is usually advisable to 
add several sections in the part marked C. 

In order to determine the conditions which are required to inject 
the necessary current into the line to provide the potential to cut off 
anode current in V 2 and so avoid deterioration of the resultant wave- 
form, the following procedure is adopted. Since the impedance of the 
line is R ohms and, at the point where the signal is injected, there are, 
in effect, two lines in parallel, one being the part A and the other the 

parts B and C, the effective input impedance is very nearly — . 



Then 



f zE 

7e' 



APPENDIX VI 



367 



where / is the required anode current in amperes in the valve V t and 
£is, say, at least 100 per cent more than the cut-off potential of the 

valve V 2 . 

It is necessary to make E at least double the potential required to 
cut off anode current in V 2 in order to preserve the steepness of the 
wave-front at the anode of K 2 . A further reason is that, due to losses 
in the part B of the line, the potential at the grid of F 3 will be reduced 
by comparison with that of V 2 . It may even prove advisable to increase 
the value of E still more. The change of anode current in the valve F, 



(a) Potent/a/ applied to the Grid of V L 



K 



~K 



if 



(l) Potential at the Anode of V L 

^Suppressed by Grid Current 

K II " 

(c) Potential at the Grid of V 2 



u U 

(d) Potential at the Grid of % 
Fig. 253 

must be /, as calculated above, and R 2 should generally be high com- 
pared with R t but it should not be so high that C 3 has not sufficient 
time to discharge before the arrival of the next unit function. The 
circuit of Fig. 252 is intended to operate with a positive unit function 
applied to the grid of V t since this will result in a negative unit function 
being produced across the input to the line. 

A point to be noted is that the anode load resistances 7? 4 and R 6 
must be of suitable values. If the time constants C b R A and C 7 R G are 
not small compared with the duration of the pulse, distortion of the 
output pulses will occur and this will appear as curvature of the sides 
of the pulses. The resistances R 4 and R t should, therefore, have as 
low a value as can be tolerated having regard to the required amplitude 
of the output signal. If, after reducing the values of these resistances, 
the pulses are still distorted, some improvement may be obtained by 
further reduction of the value of the resistances and by compensating 
the reduced output by increasing the screen potential of the valves 
V 2 and V 3 and so increasing the anode current. 

Fig. 253 shows the wave-form associated with various parts of the 
circuit of Fig. 252, from which it will be seen that pulses are produced 



3 68 



TIME BASES 



at each change of level of the input potential. The resulting positive 
pulses are suppressed by grid current in V 2 and further suppression 
is introduced at the grid of V 3 . Note that although the mark-space 
ratio is shown as unity in Fig. 253 (a), this is not a necessary condition. 
Note also that, in Fig. 253, the potential at the grid of V z has been 
shown correctly, i.e. it is delayed with respect to the potential at the 
grid of V 2 - 

The Use of a Line to Delay a Signal 

When it is particularly required to examine the commencement of 
a signal with the aid of a single-stroke time base, the signal may be fed 
to the deflector plates of the cathode ray tube via a correctly-terminated 
delay line. In this way, the necessary short interval for the time base 
to commence to operate may be allowed to elapse before the signal 
arrives at the cathode ray tube. 



APPENDIX VII 
A METHOD OF CHANGING THE PHASE OF A SINUSOID 

It is sometimes required to provide a means of varying the phase 
of a low-frequency sinusoidal potential in order to employ it, after 
squaring the wave-form, for the purpose of controlling a time base, 
a calibrating oscillator or other device. This may be accomplished by 
means of the circuit shown in Fig. 254, in which the transformer T is 




Sinusoidal 
Input 



Fig. 254. Method of changing the phase of a sinusoid 



The 






shunted by a resistance and a capacitance connected in series, 
value of — should be made approximately equal to one-tenth of the 

(|>C 

maximum value of the variable resistance R. The transformer tap 
should be in the electrical centre of the secondary winding, i.e. the 
potentials between each end of the winding and the centre tap should 
be equal. 

The operation of the circuit may be seen from an examination of 
Fig. 255, which shows the circle diagram for the circuit. The potential 
E 1 +E 2 across the secondary 
winding is shown as a horizontal 
vector and a semicircle is con- 
structed upon it because the 
condenser and resistance are 
considered to be a pure capaci- 
tance and resistance respectively 
so that the phase angle between 
the potentials across these com- 
ponents will always be 90 . The 
resistance of the transformer 

windings is neglected in the diagram. Provided no attempt is made 
to draw an appreciable current from the output, the phase-change 
26 369 




Fig. 255. Circle diagram illustrating the 

change of phase produced by the circuit 

of Fig. 254 






37° 



TIME BASES 



between the output and input potentials may be controlled within the 
limits of approximately 20 and 160 without appreciable change of 
amplitude at the output terminals. This type of phase shifter has been 
employed in Builder's frequency- dividing circuit (see Fig. 170). 

If it is required to obtain a greater degree of phase-shift, a reversing 
switch may be connected either to the primary or to the outer ends of 
the secondary winding of the transformer. This has the effect of adding 
180 to the phase-shift provided by the network. For other forms of 
phase-shifting network see also page 335. 

A phase-shifting circuit which employs a valve instead of a trans- 
former and which may therefore be used over a wide range of frequencies 
has been described by Sowerby.* 

*R. McG. Sowerby: "Electronic Circuitry," Wireless World, 1949, 55, p. 23. 



APPENDIX VIII 

THE EFFECT OF LARGE VALUES OF SHUNT CAPACITANCE 
IN THE ANODE CIRCUIT OF A VALVE 

It is interesting to note the effect of a large value of capacitance (i.e. 
large enough to cause an appreciable phase shift when the valve is 
employed to amplify an alternating potential) connected between the 
anode of a valve and one of the supply rails. The result is the same 
as is produced by the integrating circuit of Fig. 216 so far as the 
output wave-form is concerned, provided there are no distortion effects 
in the valve. The output potential is, of course, increased because of 
the amplification of the valve, provided that gR 2 is greater than unity. 
Fig. 256 shows a pentode valve, having a resistance R t shunted by a 
capacitance C 2 , in the anode circuit. In order to determine the effect of 
the condenser, it is necessary to consider C 2 as part of an integrating 



Rz 



Q rr 
r ° — Il-r~ 

Input 5 — 

Potential 



\Ri 




OHX+ 



oOut 



o/f.r- 



Fig. 256. Amplifier having a large value of capacitance C a in the anode circuit 

network in which the resistive arm comprises R 2 in parallel with the 
internal anode-cathode impedance of the valve. The effective input 
potential to the network is given by: 



*-'GJ&K 



where g is the slope of the valve characteristic in amperes per volt, in 
the operating condition, 

R z the value of the anode load resistance, in ohms, 

R a the value of the internal anode-cathode impedance of the 
valve, in ohms, 

e g the alternating component of the potential applied to the 
grid (in the case of a sinusoidal input this potential is in 
R.M.S. volts and for a saw-tooth, square or triangular 
wave-form the peak potential is used), and 

E is in R.M.S. or peak volts exactly as is employed for e g . 

26* 37 * 



372 



TIME BASES 



For a sinusoidal input, the phase-change is 

6 = tan^wT, 
where is the phase lag in degrees, 

co is 2-/ 

/ the frequency, and 



T = C< 



/ R 2 R a \ 



in seconds. 



The ratio of the output potential to that of the input for a sinusoidal 
input potential is given by 

£ = cos 0, 

where e is the output potential and is the phase-shift in degrees. 

In the case of Fig. 256, if the input to the grid of the valve V l is 
sinusoidal and no distortion takes place in the valve, the phase-shift 
and attenuation may be obtained from the curves given in Figs. 229 
and 230. 

It will thus be appreciated that, for a given value of <aT, there is 

e 

only one value of the phase-angle and of the attenuation — regardless 

E 

of the value of R 2 which has already been taken into account in 

equation (1) 

When the valve V 1 is a pentode, the behaviour of the circuit may be 
explained as follows: 

The pentode valve forces a change of current gSe s through the 
anode load provided the impedance of the anode load is not suffi- 
cient to cause the anode potential to fall below the knee of the 
curve and the grid potential change 8e g does not become sufficiently 
negative to cut off the anode current. The pentode forces a current 
through the circuit because the internal impedance of the valve and, 
hence, the impedance of the source supplying current to the network 
is high with respect to that of the network. 

The network is therefore a current-fed network and, from this point 
of view, C 2 and R 2 mav be considered as being connected in parallel. 
See Fig. 220 {e) and Table XVI. 

The Use of a Multi-stage Integrating Network 

It is sometimes desired to obtain a greater amount of phase-shift 
than can be derived from a single-stage network. A single network gives 
a maximum phase-change of 90 and, in practice, it is limited to 
less than this figure by consideration of the attenuation. For a given 
phase -shift the output may be raised by one or other of the following 
methods: 



APPENDIX VIII 



373 



(a) The potential E may be raised by increasing g and so providing 
a larger change of anode current. 

(b) By an increase in the value of R 2 , the value of C 2 being reduced 

/ R R \ 
appropriately in order to keep the value of C 2 ( 2 a 1 

constant, or ^ 2 + ^ 

(c) By the use of two or more networks instead of only one. In 
this case it is usual to make the phase-shift 6 for each network 

equal to where N is the number of networks employed, and 8' 

N 

is the total phase change required. 

Fig. 257 shows a circuit employing three separate networks and the 
object is to ensure that each network is operated efficiently in the sense 
that the phase-change is large in proportion to the attenuation. If a 
total phase-change which is appreciably in excess of 6o° is required, 
it is necessary to employ two or more networks in tandem if the output 
potential is not to be seriously reduced. 

For example, with a single network, cos 6=0-5 when 0=6o° and 
the output is therefore 0-5 of the input potential. If, on the other 
hand, three networks are employed, each providing a phase-shift 
equal to 20 , it emerges that cos 0=0*939 anc * t ^ ie out P ut potential is 
°"939 s or 0*829 of the input. 

In order to avoid undue loading of the capacitance arm of each 
network by the following network, the impedance of successive net- 



OH.T.+ 



O0ut 



oH.T- 



Fig. 257. Amplifier circuit having three phase-changing networks 

works should be increased by a factor of, say, 10. In order to obtain 
the maximum benefit, the phase-shift for each network should be equal 
and, hence, the time constant is made the same in each case. 
In the circuit of Fig. 257 the time constants are thus ; 




Input 
Potential >R-i 



T, = C 



and 



.#2 + 

-* 3 = ^3 "3 
1 4 = C d i\4 



-..(3) 
.-.(4) 



I 1 



374 TIME BASES 

and the impedances of the three networks are : 



APPENDIX VIII 



375 






i 

OiC*; 



Z* = ^ Ji +»*C,*K,* f 



^4 = 



7 



i + o>*e 4 2 tf 4 2 



It will be seen that if 



R = I0 R = IO o(~-?-°-^ 

8 V^ 2 + R.) 



and 



C 4 = - 3 = ^ 2 



10 100 

then Z 4 = ioZ 3 = iooZ 2 

and r 2 = r, = r 4 . 

The above equations are applicable, whether the valve is a triode or 
a pentode but, where R a is greater than, say, io R z , the term R % may 



be substituted for 



R%Rg. 

RzTR a 



without serious error. 



In order to determine the values of the component parts of the 
circuit, it is first necessary to determine the minimum permissible 
output amplitude e and to decide upon the amount of phase-shift d' 

required. If the resultant value of -% is greater than cos 0' more than one 
network will be required. The number of networks required may be 

6 

found by trial and error from the equations Q' =N0 and -% =(cos 0) N 

using Figs. 229 and 230. 

The design may be considered from several possible points of view, 
e.g. C 4 must not be smaller than the input capacitance of the valve 
which is to follow the final network. If a large output potential is 
required, one may commence by drawing the load diagram (ellipse) 
on the anode characteristic of the valve V l and determine the necessary 
value of e g . If, however, e g is fixed, one may be able to choose a 
particular valve and to employ a suitable value of the supply potential. 
Having decided upon the number of networks to be employed, obtain 
the time constant required for each network from Fig. 229. The next 
operation is to determine the value of R 2 , which will provide the 
assumed potential E. This is determined with the aid of equation (1) 
if e s is given. Since R 2 is now known, the value of C 2 is obtainable 
from equation (2) and the values of the remaining components may 



be obtained from equations (8) and (9). The valve F L in Fig. 257 
should, of course, be biased to the centre of the characteristic and the 
input potential should be sufficiently small to avoid overloading the 
valve. 

As already stated, in order to reduce loading effects, the impedances 
of the networks are increased progressively by a factor of 10, but this 
does not entirely eliminate the effect. In order to compensate for any 
small remaining effect, and for inaccuracy of the values of the com- 
ponents, one of the resistances or one of the capacitances may be made 
capable of a small amount of variation in order to obtain the precise 
phase-shift required. 

If the input potential is not sinusoidal, the output wave-form will 
differ from that of the input and the effect of one network may be 
estimated by examination of Figs. 217 to 219. If more than one network 
is employed, the wave-form change is more complex, but the general 
trend is to eliminate rapid changes, eventually leaving a pure sinusoid 
if the process is carried sufficiently fur. 

The networks described above may be used for initiating, say, a time 
base and a calibrating oscillator in such a way that the oscillator is 
made to commence operation before the time base trace commences. 
The procedure to be adopted is that the potentials across two parts 
of the network are taken to the control grids of two separate squaring 
valves (see page 351) in order to produce two square waves whose 
phase is slightly different, i.e. one square wave commences just before 
the other. The earlier square wave is employed to operate the cali- 
brating oscillator (see page 143) and the other to control the time 
base. 

Another use for the networks described above is for the generation 
of oscillations.* If the output of the network is connected to the grid 
of the valve to whose anode the network is connected, the circuit will 
oscillate provided the overall gain is greater than unity. The sinusoidal 
input is, of course, removed. The frequency of oscillation will be that 
which will result in a total phase-shift through the network of 180 
and, therefore, at least two networks, and, in practice, to avoid too 
great an attenuation, at least three networks are necessary. 

If one of the grid leak resistances in a multivibrator is replaced by 
a parallel tuned circuit, the output at each anode will be sinusoidal. 
Similarly, a sinusoidal output can be obtained if the two pairs of anode- 
grid connections, e.g. C X R 2 and C 2 R 1 in Fig. 15 are each replaced by 
networks which provide a phase change of +90° and — 90 respectively 
at some desired output frequency. The phase change in each valve, at 

*G. Wiltoner and F. Tihelka: "A Phase-Shift Oscillator with Wide-range Tuning," 
Proc. I.R.E. t 1948, 86, p. 1096; C. L. Longmire: "An R. C. Circuit giving over-unity 
gain," Tele-Tech., Part I, 1947, p. 40. 



376 TIME BASES 

low frequencies, is 180 , a total of 360 or %t. and the phase changes in 
the two networks cancel. Such an arrangement is a very useful method 
of producing a sinusoidal source of low frequency. 

Three-phase sinusoidal oscillators may be made by using three valves 
having networks which provide a phase change of 6o° between each 
valve, the output being taken from the three anodes. In this case, the 
total phase change is 37c in the valves and - for the three networks, a 
total of z~ or 4- depending upon whether the output from each network 
leads or lags with respect to its input potential. 

Both these circuits employing phase -changing networks have the 
advantage that the load upon the supply rail can be made constant so 
that, even if the generated frequency is only one or two cycles per 
second, the final smoothing condenser in the H.T, supply system need 
only be made sufficiently large to reduce the rail hum to suitable 
proportions. This remark applies to all circuits in which feed-back is 
applied only round the complete loop, in which valves of one type 
only are used and each valve has equal gain, the same circuit values 
and the same values of applied potential. When these conditions are 
fulfilled, the sum of the alternating components of the anode currents 
is at all times zero; the same remark applies to the screen grid currents. 






BIBLIOGRAPHY 

The following books and papers contain useful information. They may, 
in general, be seen at the Patent Office and at certain Libraries. Copies of 
Patent Specifications may be obtained, price zs. each (2s. id. abroad), from 
the Patent Office, 25 Southampton Buildings, London, W.C.2. 

A full description of the facilities offered by the Patent Office Library is 
given in Electronic Engineering. 

BOOKS 

S. G. Starling. Electricity and Magnetism. Longmans, Green. 

W. B. Coulthard. Transients in Electrical Circuits. Pitman. 

E. Williams. Thermionic Valve Circuits. Pitman. 

A. C. Cossor, Limited. The Cossor Double Beam Oscillograph. 

Massachusetts Institute of Technology. Waveforms. McGraw Hill. 

Massachusetts Institute of Technology. Electronic Time Measure- 
ments. McGraw Hill. 

A. A. Andronow and C. E. Chaikin. Theory of Oscillations. Princeton 
University Press. 

G. F. J. Garlick. Luminescent Materials. Oxford University Press. 

A. W. Keen. Principles of Television Reception. Pitman. 

PAPERS 

R. Neumann. The Patent Office Library. Electronic Engineering. 1949, 

21, p- 52. 
M. L. JoFEH. An Operational Treatment of the Design of Electromagnetic 

Time-Base Amplifiers. Journal I.E.E. 1939, 85, p. 400. 
R. A. Monfort and F. J. Somers. Measurement of the Slope and Duration 

of Television Synchronizing Impulses. R.C.A. Reviciv. 1942, 6, 

p. 370. 

D. C. Espley and G. W. Edwards. Television Receivers. Journal Tele- 

vision Soc. March 1938, p. 363. 
A. Preisman. Graphics of Non-linear Circuits. R.C.A. Review. 1937, 

2 (part 1), p. 124 and (part 2) p. 240. 
H. Nyquist. Regeneration Theory. Bell Sys. Tech. Journ. 1932, 11, p. 126. 

E. Peterson, J. G. Kreer and L. A. Ware. Regeneration Theory and 

Experiment. Bell Sys. Tech. Journ. 1934, 13, p. 680. 
A. Preisman. Some Notes on Video Amplifier Designs. R.C.A. Review. 

1938, 2, p. 421. 
S. W. Seeley and C. N. Kimball. Analysis and Design of Video Amplifiers. 

R.C.A. Review. 1937, 2, p. 171. 

F. Vecchiacchi. Multivibrateur Dissymetrique. L'onde Electrique. May 

1932, p. 16. 

377 



378 



TIME BASES 



S. R. Jordan. The Controlled Transitron Oscillator. Electronics. 1942, 

15, p. 42. 
H. J. Reich. The Use of Vacuum Tubes as Variable Impedance Elements. 

Proc. LR.E. 1942, 30, p. 288. 
N. V. Kipping. Investigations on some Valve Circuits with the Cathode 

Ray Oscillograph. Wireless World. 1923, 12, p. 827. 
Investigations with the Cathode Ray Oscillograph. Wireless World. 

1923. 13 > P- 309- 
A Method for accurate Frequency Calibration. Wireless World. 1924, 

13, p. 639. 

A Practical Demonstration of some Applications of the Cathode Ray 

Oscillograph. Wireless World. 1924, 13, pp. 705 and 805. 

11. Moss, Cathode Ray Tube Traces. Electronic Engineering, June, July, 
August and September 1944 (Lissajous Figures). December 1944 and 
January 1945 (Straight Line Time Bases). October 1945 (Circular 
and Spiral Time Bases). April, June and August 1946 (Complex 
Wave-forms). December 1946 (Modulated Waves). 

K. A. Pullen. The Cathode Coupled Amplifier. Proc. LR.E. 1946, 34, p. 403. 

T. W. Winternitz. A Variation on the Gain Formula for Feed-back 
Amplifiers for a Certain Driving- impedance Configuration. Proc. 
LR.E. 1946, 34, p. 639. 

P. G. Sulzer. Cathode-coupled Negative-Resistance Circuit. Proc. LR.E. 

1948, 36, p. 1034. 

J. R. Tillman. Electronic Negative Resistors. Wireless Engineer. 1945, 

22, p. 17. 
C. F. Brockelsby. Negative Feed-back Amplifiers — Conditions for Maximal 

Flatness. Wireless Engineer. 1949, 26, p. 43. 
A. W. Keen. The Integration Method of Linearizing Exporcntial Wave- 
forms. Journ. Brit. LR.E. 1949, % P- 4 : 4- 
A. M. Skellett. Use of Secondary Electron Emission to obtain Trigger 

or Relay Action. Journ. Applied Physics. 1942, 13, p. 519. 
J. McG. Sowerby. The See-saw Circuit Again. Wireless World. 1948, 54, 

p. 447. 
E. L. Clark. Automatic Frequency-phase Control of Television Sweep 

Circuit. Proc. LR.E. 1949, 37, p. 497. 
S. A. Knight. Cathode Ray Oscilloscope. Wireless World. 1948, 54, p. 432. 
M. Silver. High Ratio Multivibrator Frequency Divider. Radio and 

Television News. July 1949, p. 7. 
P. G. Sulzer. A Wide-range Saw-tooth Generator. Rev. Set. Insts. 1949, 

20, p. 78. 
G. L. Hamburger. Graphical Analysis of Diode Circuits. Wireless Engineer, 

1949, 26, p. 147. 

W. T. Cocking. Blocking Oscillators. Wireless World. 1949, 55, p. 230. 
H. Mayr. Feed-back Amplifier Design. Wireless Engineer. 1949, 26, p. 297. 
V. H. Attree. A Slow-sweep Time Base. Journ. Sci. Insts. 1949, 26, p. 257. 
T. W. R. East and A. F. Standing. A Circular Time Base Frequency 
Comparator. Journ. Sci. Insts. 1949, 26, p. 236. 



INDEX 



Each letter-section of the index is divided into two portions for convenience. 
The first portion is devoted to subjects and the second to authors and inventors. 



Abraham and Bloch's Multivibrator, 28 
A.C. Operation of Cathode Ray Tube, 
288 

time bases, 1 1 
Actinic efficiency of screen, 294 
After-glow of screen, 294 
Aluminized screen, 295 
Amplification of balanced potentials, 218 
Amplifiers, 191 

anode-coupled, 223 

cathode-coupled, 212 

design of circuits, 203 

distortion in, 191, 205, 215, 224 

feed-back, 28, 196, 346 

galvanometer, 84 

input impedance, 201, 202, 207 

in series, 206 

integrating, 141 

negative feed-back, 129, 131, 158, 196, 
340 

output impedance, 134, 200, 202, 206 

phase reversing, 192 

push-pull, 192, 210 

shaping effect of, 207 
Angle of deflection, 142 
Anode capacitance, 192, 371 

cathode impedance, 59 

coupling, 223 

current limitation, 22, 310 

follower, 202 
Anti-astigmatism circuit, 278 

trapezium tube, 276 
Appleton, Herd and Watson Watt's time 

base, 89 
Astigmatism, 278, 286 
Asymmetric distortion, 276 
Attenuation in networks, 323, 335, 346 
Attree's time base, 171 

Abraham, H., 28 

Allen, A., 75 

Andrew, V. J., 244 

Animaxander, 1 

Anson, H. St. G., 15 

Appleton, E. V., 16, 89 

Ardenne, M. von, 17, 104, 142, 192, 261 

Arman, L. T., 80 

Attree, V., 171 



Back-lash, 81 

Barrel distortion, 154, 283 

B.B.C. raster, 7, 154 



Bcale and Stansfield's integrating am- 
plifier, 140 
Beam current, 7, 284 

magnetic field due to, 290 

velocity, 275 
Beat -frequency oscillator, 155 
Bedford and Steven's method of lineari- 
zation, 114 
Bedford's method of A.C. operation of 
cathode ray tube, 288 

method of linearization, 115 
Benjamin's blocking oscillator counter, 

257 
Black-out signal, 7, 41, 273 
Black's time base, 85 
Blocking oscillator, 89 

oscillator counter, 257 

oscillator pulse generator, 362 
Blumlein's electromagnetic time base, 95 

method of linearization, 137 

thyratron time base, 25 
Bootstrap circuit, 135 
Bottoming, 1S2 
Brightness modulation, 273 

modulation, avoidance of, 274 

trace, 7, 273 
Bucking wires, 262 
Buffer valve, 132 
Builder's frequency divider, 244 
Burning of screen, 53, 266 

Bainbridge-Bell, L. H., 103 

Beale, E. S. L,, 141 

Bedford, A. V., 95, 137, 240 

Bedford, L. H., 17, 20, 108, 114, 115, 

195, 261, 275, 288 
Berg, W. F., 298 
Bijl, H. J. van der, 260 
Black, D. H., 85 
Black, H. S., 128 
Bloch, E., 28 
Blumlcin, A. D., 25, 95, 97, 137, 171. 

213, 219, 267. 34i. 364 
Bode, H. W., 204 
Bowman-Manifold, M., 137 
Braun, F-, 259 
Briggs, B. H., 106 
Brunetti, C, 57 
Builder, G., 237, 243 
Bull, E. W., 137 
Bullen, T. G., 104 
Bush, V., 345 
Busse, E., 130 
Butler, O. I., 237 



379 



3 8o 



TIME BASES 



INDEX 



38i 






Calibrating oscillator, 143 
Calibration of time base, 8, 142 
Capacitance, anode, 40, 371 

cathode, 21, 129, 133 

grid, 133, 341, 344 

minimum in thyratron time base, 314 

stray, 8, 23, 30, 39, 133, 166, 183, 342 
Capacitive time bases, 14 
Carpenter's anode-coupled push-pull 

amplifier, 223 
Cathode capacitance, 21, 129, 133 

capacitance charging, 21, 129, 133 

controlled frequency divider, 250 

controlled sanatron, 187 

coupled amplifier, 212 

coupled amplifier, gain control, 216 

coupled phant, 173 

coupled trant, 70 

follower, 39, 129, 131, 218, 293 

follower probe, 293 

heater potential, 23, 41 

load, 129 

ray beam, 261, 269, 275, 290 

ray focusing, 260 

ray tube, 104, 259 

ray tube, A.C. operation, 288 

ray tube, actinic efficiency, 294 

ray tube, afterglow, 294 

ray tube, anti-trapezium, 262, 276 

ray tube, cross-talk, 29 r 

ray tube, distortion in, 275 

ray tube, double beam, 26 1 

ray tube, effect of magnetic fields, 289 

ray tube, gas-focused, 260 

ray tube, lens system, 260 

ray tube, operating potentials, 263 

ray tube, phosphors, 297 

ray tube, photographic data, 298 

ray tube, photographic errors, 303 

ray tube, polar co-ordinate, 104 

ray tube, power supplies, 263 

ray tube, screens, 294 

ray tube, secondary emission in, 294 

ray tube, shift potentials for, 285 

ray tube, stray fields, 290 

ray tube, use of, 289 

ray tube, writing speed, 301 

rays, 259 

resistance, 39, 129, 213 
Chance's trigger circuit, 84 
Charge on a condenser, 14, 241, 305 

percentage, 112 
Charging, cathode capacitance, 21, 129, 

133 
characteristic, 305 
current, 112, 305, 314 
device, 1 6, 17 
diode, 16 
grid current, 89 
pentode, 17 
period, 305 
rate, 24, 305 
resistance, 16 
time, 24, 33, 305 



Chopper, 351 

Circuit linearity, 114, 124 

switching, 14 
Circular time base, It, 100 
Clare's push-pull amplifier, 217 
Clayden's frequency divider, 248 
Clepsydra, 1 
Clipper, 351 
Cockroft and Walton's voltage-doubler, 

264 
Coils, deflection, 94, 137, 269 
Colour, screen, 294 

Compensating networks, 97, 117, 138,334 
Computing networks, 345 
Concept of virtual earth, 204 
Condenser, blocking, 128 

charge on, 14, 24, 305 
Constant-current device, 17, 129, 158 
Control ratio, 20, 311 
Cossor cathode ray tubes, 262, 295 
Counter, blocking oscillator, 257 
Counting, 256 
Coupling, anode, 223 

cathode, 213 

condensers, 12S 

screen, 35, 64, 178 

screen to suppressor, 57 
Cross-talk, 291 
Current, charging, 112, 305, 314 

discharge, 18, 313 

feed-back, 129, 196 

leakage, 18, 22, 128, 129, 166 

minimum, 313 

percentage charge, 113 

regulating circuit, 156, 159 
Curvature of charging characteristic, 14, 
16, 112, 305 

of screen, 141, 303 
Cyclogram, 4, 50, 317 



Campbell-Swinton, A., 259 
Carlson, W. L., 93 
Carpenter, R. E. H., 210, 223 
Carslaw, H. S., 332 
Cattanes, E., 303 
Chance, B., 84, 360 
Chakravarti, S. P., 55 
Clare, J. D., 217 
Clarke, A. C., 114 
Clayden, R. T., 248 
Cocking, W. T., 210 
Cockroft, J. D., 264 
Colebrook, F. M., 84, 158, 340 
Corbeiller, P. le, 55 
Ctesibius, 1 
Currah, L. E,, 35 



Dark trace cathode ray tube, 298 
D.C. amplifier, 84 
time bases, 13 






Deflection angle, 142 

coils, 137, 269 

defocusing, 276, 283, 287 

distortion, 112, 276, 279 

electromagnetic, 272 

electrostatic, 269 

errors, 112, 271, 276, 279, 343 

magnetic, 269, 289 

methods, 268 

plates, 268 

push-pull, 190 

sensitivity, 102, 269 
Defocusing, deflection, 276, 283, 287 

due to stray fields, 289 
Deionization time, 309, 322 
Delay circuit, 173 

line, 45, 364, 368 
Den Hartog and Muller's time base, 43 
Developers, Photographic, 302 
Differentiating circuit, 339 

networks, 323, 348 
Diode charging, 16 

discharging, 89 

limiter, 353 
Discharge device, 9 

diode, 89 

rate limitation, 22, 39 

time, 24, 192 

valve anode current, 22 
Displacement error, 112 
Distortion, amplifier, 191, 205, 215, 224 

astigmatism, 278 

asymmetric, 276 

barrel, 154, 283 

fly-back, 192 

in cathode ray tube, 275, 279 

measurement, 142, 279 

origin, 142, 260 

pin-cushion, 154, 283 

trace, 112, 279, 343 

trapezium, 154, 276 

wave-front, 343 
Divider, 49, 173, 244, 248 
Division of frequency, 242 
Dodds' time base, 229 
Double-beam cathode ray tube, 8, 261 
Dye's time base, 101 

Debski, J. M., 171 
Den Hartog, H., 41, 43 
Dodds, E. M., 229, 303 
Dow, W. G., 316 
Dowling, J. J., 104 
Duke, V. J., 156 
Dunham, C. R., 25 
Dye, D. W., 101 



Eccles-Jordon trigger circuit, 76 
Effective H.T. potential, 130, 132, 343 
Efficiency, actinic, 295 

luminescent, 295 

of a time base, 16 



Electromagnetic deflection, 269, 272 

screen, 291 

time base, 95 
Electrostatic deflection, 268 

screen, 23, 41 
Elliptical time base, 11, 100 
E.M.I, cathode my tubes, 296 
Emission peaks of screens, 299 

secondary, 85, 294 
Emulsions, photographic, 302 

sensitivity peaks, 299 
Engines, testing internal combustion, 234 
Equivalent impedance, 129 

potential, 130, 132, 343 
Errors, deflectional, 112 

displacement, 112 

linearity, 112, 142, 162, 279, 343 

phase-shift, 192 

photographic, 303 

trace velocity, 112 
Evans' trigger circuit, 80 
Expanded time base, 41, 1 1 1, 221 
Exponential charge, 14, 323 
Exponents, table of, 306 
Exposure times, 300 
Externally operated time bases, 52 
Extinction current, 313 

potential, 16 

Eccles, W. H„ 76 
Evans, G. T., 80 



Fant, 172 

cathode-coupled, 173 

screen-coupled, 178 
Faudell's method of linearization, 139 
Feed-back amplifiers, 28, 196, 203, 349, 

375 

amplifiers, design of, 203 

amplifiers, input impedance, 201, 208 

amplifiers, output impedance, 200, 206 

amplifiers, shaping effect of, 207 

current, 129, 196 

linearization, 128, 158, 340 

negative, 129, 131, 158, 196, 340 

positive, 28, 55, 141, 145, 147 

voltage, 133, 198 
Fleming -Williams' double beam cathode 

ray tube, 261 
Flicker, television, 7 
Flip-flop, 56, 64 

cathode-coupled, 70 

early reinitiation, 75 

screen-coupled, 64 

single-valve, 64 

transitron, 64, 70 

two- valve, 72 
Fly-back, 7, 25, 171, 192 

correction, 192 

duration, 25, 39, 40. IQ 2 

linearization, 41, 192 

trace suppression, 41, 53 

velocity, 39, 4 J > l 7 I > 1Q 2 



382 



TIME BASES 



Fluorescent screens, 294, 297 
Flux measurement, 150 

orientation, 291 

variation, 150 
Focusing, 260, 267 
Fog, photographic, 300 
Frame, 154 
Frequency divider, 49, 173, 244 

divider, sanatron, 250 

divider, santrig, 251 

divider, scalariform, 254 

division, 242 

division from a pulse, 248 

division from a sinusoid, 243 

repetition, 25 
Friihauf's time base, 36 
Function, step, 207, 332, 342, 364 

unit, 227, 332, 342, 364 

Famsworth, P. T., 125 

Faudell, C. L., 28, 91, 92, 93, 95, 137, 

139, 240, 341 
Faulkner, H., 80 
Fleming, A., 259 
Fleming-Williams, B. C, 8, 60, 72, 178, 

262, 277. 356 
Folkerts, H. P., 298 
Friihauf, G., 36 



Gas-iilled relay, 20 

triode, 20, 230, 309 

triode characteristics, 309 

triode time base, 19, 121, 230, 309 
Gate valve, 44 

Godeck's radial time base, 106 
Graticule, use of, 302 
Grid bias potential, 19, 32, 65, 70, 312 

capacitance, 21, 133, 341, 344 

-controlled sanatron, 187 

current charging, 89 

emission, 22 

limitation, 22, 310, 312, 354 

potential, positive, 32, 70 

protection, 22 

temperature, 22 

Garlick, G. F. J., 298 
C J it tins, J. P., 298 
Gleason, C. H., 264 
Gobau, G., 106 
Goddard, M. J., 44 
Godeck, J, E., 106 
Goldmuntz, L. A., 356 
Gordon, R. L., 217 
Grosdoff, I. E., 256 
Groves, W. R., 129 



Hawkins' method of linearization, 117 
Heater-cathode hum, 23, 41 

-cathode potential, 23 

transformer, 23, 41 



Heaviside unit function, 207, 332, 342, 364 
Hollman effect, 262, 271 
Hollman's three-phase tube, 263 
Holmes, Carlson and Tolsons" time base, 

93 
Hum, 23, 41 

Hawkins, G. F., 117 
Head, R. B., 261 
Heaviside, O., 342, 364 
Henderson, S. T., 298 
Hendry, N., 298 
Hercock, R. J., 298 
Herd, J. F., 16, 89, 103 
Hcrold, E. W., 55 
Hirshman, C. L., 228 
Hollman, H, E., 263, 272 
Holmes, R. S., 93 
Hopkinson, R. G., 297, 298 
Hosking, J., 238 
Howe, G. W. O., 28, 32 
Howell, W. D., 106 
Hull, A. W , 19, 85 



Image brightness, 6, 109, 273 

repeating, 300 

television, 7, 108 

two-dimensional, 6 
Impedance equivalent, 129 

input, 198, 20 r, 202, 207 

of balanced amplifier, 201 

of cathode follower, 198 

of feed-back amplifier, 206 

of supply circuits, 23, 376 

operators, 210 

output, 1 8, 134, 159, 200, 202, 206 

slope, 17 
Inductive linearization, 124 

pulse generator, 360 

time base, 15, 95 
Inherent locking, 44 
Integrating amplifier, 141 

circuits, 108, 158, 207, 323, 340, 371 

networks, 1 17, 323, 371 
Integrator, Miller, 158, 340 
Ionic recombination, 309, 320 
Ionization, 15, 309 

residual, 309 
Interlaced scanning, 7, 154, 238 



Jenkins' method of avoiding brightness 
modulation, 274 

method of linearization, 116 
Jitter, 34, 76 

prevention of, 44, 76, 238, 240 

Jaeger, J. C, 332 
Jenkins, J. W., 116, 274 
Jofeh, M. L., 216 
Johnson, J. B., 260 



Johnson, L. R, J., 341 
Johnston, D. L., 219 
Johnstone Stoney, G., 259 
Jordan, F, W., 76 



INDEX 



383 



Lewis, W. B., 256 
Lilienfeld, E., n, 100 
Longmire, C. L., 375 



Keen's method of linearization, 122 

time base, 41 
Kobayashi's time base, 91 

Kallmann, H. E., 364 
Kaufmann, W., 259 
Keen, A. W., 41, 122, 239 
Kilburn, T., 203, 346 
Kimball, C. N,, 150 
King, P. G. R-, 298 
Klemperer, O., 261 
Kobavashi, M.,- 91 
Koch^ W. R., 266 
Koga, I., 244 
Krauss, H. L., 131, 356 



Law's pulsed calibrating oscillator, 147 
Leakage currents, 18, 22, 128, 129, 166 
Leaky condensers, effect of, 18, 22, 128 
Lens system of cathode ray tube, 260, 

268, 276 
Light, wave-length of, 299 
Lilienfeld's time base, 11, 100 
Limitation, anode current, 22, 310 

grid current, 22, 310, 312, 354 
Limiter, 30, 351, 353 

valve, 30, 351 , 353 
Linear shaping circuits, 328 

time base, 13 
Linearity error, 112, 142, 162, 279, 343 

measurement, 142, 279 

photographic, 303 

raster, 153 

trace, 18, 112, 142, 162, 279 
Linearization, 112 

auxiliary time constant, 117 

feed-back, 128, 158, 340 

inductive, 25, 124 

valve curvature, 114 
Linearizing circuits, 114 

networks, 117 
Lines, delay, 45, 364 
Locking, inherent, 44 

suppressor, 68 
Long-tailed pair, 216 
Low potential regime, 318 
Luminescent efficiency of screen, 295 

Laws, C. A M 147 
Le Corbeiller, P., 55 
Leslie, C. B,, 256 
Leverenz, H. W,, 261, 297 
Levy, L., ;6i 
Lewer, S. K,, 25 
Lewis, H. M., 60 



Mac Kay's template generator, 363 

time base, 171 
Magnetic deflection, 268, 272, 289 

deflection, direction of, 269, 291 

interference, 289 

screen, 291 
Marconi-E.M.I. television raster, 7, 154 
Mark-space ratio, 30, 356, 360 
Mazda cathode ray tubes, 296 
Measurement of deflection distortion, 
142, 279 

of flux variations, 130 

of linearity, 142, 279 

of spot size, 283 

of trace linearity, 1 42, 279 

of trace velocity, 142 
Miller-capacitance, 158, 340 

-capacitance, time base, 158 

effect, 158, 340 

integrator, 340 

time base, 158 
Modulation, brightness, 273 

brightness, unwanted, 274 
Moss's time base, 44 
Milliard cathode ray tubes, 297 
Multiar, 356 

as time modulator, 360 
Multiple time base, 105 
Multivibrator, 28, 356 

as square-wave generator, 28 

standard frequency generator, 28 

frequency control, 34 

high-frequency, 28 

jitter, 34 

mark-space ratio, 3 1 

period, 30, 32 

pulse generator, 28, 356 

repetition frequency, 31 

synchronization, 34 

wave-form, 29 
Mu-metal screen, 291 

MacKay, D. M., 171, 363 

MacLeod, K. C, 267 

Manifold, M., Bowman-, 137 

Mansford, H. L., 219 

Marchand, N., 323 

Mark, J. van der, 130 

McMullen, D., 41 

Miller, J. M-, 158, 34° 

Mitchell, C. J., 158, 34° 

Monfort, R. A., 240 

Moody, N. F., 114, 158, 171. l8l < l88 > 

25°. 34 1 . 357 
Moss, H., 44, 261, 294. 3°3 
Moullin, E. B., 264 
Muller, F. A., 41, 43 
Myers, L. M., 261 



3«4 



TIME BASES 



Negative feed-back, 129, 131, 158, 196, 

340 

resistance, 55, 59, 70, 312 

slope, 70, 356 
Neon tube, 15 

tube time base, 15 
Networks, attenuation in, 323, 335, 346 

compensating, 97, 117, 138, 334 

current fed, 324, 327, 332, 364 

differentiating, 323, 348 

electrical computing, 345 

high impedance, 207, 324, 332, 348 

integrating, 323, 348, 371 

linearizing, 117 

low impedance, 207, 327, 332 

phase-shift in, 335, 371 

potential fed, 207, 324, 332, 348 

response of, 207, 323, 348, 364, 371 

shaping effect of, 207, 332, 348, 364, 

371 

Newsam's method of linearization, 131 

Non-linearity, 112, 162, 279 

Norton and Groves' method of lineariza- 
tion, 129 

Nagy, P., 44 
Nethercot, W., 298 
Newsam, B., 131, 137, 171 
Norton, F. R., 129 



Occupancy ratio, 186 
Operators, impedance, 210 

transfer, 210 
Origin distortion, 142, 260 
Oscillator, beat frequency, 155 

blocking, 89, 257, 362 

calibrating, 143 

crystal calibrating, 147 

pulsed, 146 

quenched, 145 

relaxation, 56 

resistance-capacitance, 375 

square-wave, 28, 355 

squegging, 89 

three-phase, 376 

transitron, 56 
Output impedance, 18, 23, 134, 159, 200, 
202, 206 



Parabolic wave-form, 119, 335 
Peak emission of cathode ray tube 
screens, 299 

potential measurements, 79, 270 

sensitivity of emulsions, 299 
Pentode characteristic, 17, 354 

charging, 17 

limiter, 354 
Percentage charge, 112 

fall in current, 113, 130 

fall in spot velocity, 113 



Phantastron (Phant), 172 

cathode-coupled, 173 

screen-coupled, 178 
Phase change in networks, 103, 106, 335, 

369 

changing circuits, 324, 335, 348, 369 

reversing amplifier, 1 92 

-shift error, 192 

-shift in network, 335, 369, 371 
Phosphors for fluorescent screens, 297 
Photographic data, 298 

developers, 299, 300, 302 

emulsions, 299, 302 

exposure times, 300 

fog, 300 

speeds, 299 

use of graticule, 302 
Photography of traces, 298 
Pin-cushion distortion, 154, 283 
Plan -position indicator (P.P.I,)? IQ 6 
Polar co-ordinate oscillograph, 104 

co-ordinate time base, 100 
Positive feed-back, 18, 55, 141, 145, 147 
Potential efficiency, 16 

equivalent, 130, 132, 343 

extinction, 16 

for cathode ray tube, 263 

multiplication, 264 

striking, 15, 309 
Potter's time base, 47 
Power supplies, 23, 263 

supplies, impedance of, 23, 59, 375 
Pressure-volume measurements, 11, 234 
Puckle's anti-astigmatism circuit, 278 

time base, 37 
Pulse generator, blocking oscillator, 362 

generator, delay line, 364 

generator, inductive, 360 

generator, multivibrator, 356 
Pulsed calibrating oscillator, 145 
Push-pull amplifiers, 190 

deflection, 190 

Pearson, S. O., 15 

Percival, W. S., 92, 137, 364 

Piggott, L, S., 260 

Pliny, 259 

Plucker, J., 259 

Pol, B., van der, 56 

Potter, J. L., 47, 49, 53 

Pout, H. W., 114 

Puclde, O. S., 17, 37, 43, 44, 5°, 108. 210, 

260, 275, 278 
Puleston, R., 298 



Quenched oscillator, 145 



Radial time base, 106 
Raster, 7, 153, 279 
linearity, 153, 279 



INDEX 



385 



Ratcheting, in 

Rate of rise or fall of potential, 14, 24, 40, 
160, 163, 175, 183 

of run-down, 175, 183 
Relaxation oscillators, 56 
Repetition frequency, 25 
Resistance-capacitance oscillator, 375 

charging, 14 

negative, 55, 59, 70, 312 
Return trace, 7, 25, 171, 192, 309 
Rotation of trace, direction of, 13 
Run-down, 158, 172, 184 

delayed, 171 

Randall, W. F., 291 
Regener, V. H., 77 
Reich, H. J., 55, 60, 77 
Richards, F. A., 298 
Ritson, F. J. Ui, 203, 345, 346 
Roberts, N. F., 237 
Roschansky, D., 259 



Sanatron, 181 

cathode-controlled, 187, 250 

frequency divider, 250 

grid-controlled, 187 

screen -controlled, 186 

suppressor grid-controlled, 181 

time base, 181 
Santrig frequency divider, 251 
Saturation of screen, 7, 294 
Saw-tooth, 7 

time base, 13 
Scalar! form time base, 243, 254 

wave, 365 
Scanning, television, 7, 154 
Schade's voltage-multiplier, 266 
Screen, actinic efficiency, 294 

afterglow, 294 

burning, 53, 266 

colour, 295 

-controlled sanatron, 186 

-coupled phant, 178 1 

-coupled trant, 64 

curvature, 142, 303 

emission peak, 299 

fluorescent, 294 

luminescent efficiency, 295 

mu-metal, 291 

phosphors for, 297 

saturation, 7, 294 

-suppressor coupling, 55 
Screened transformer, 23, 41 
Schmitt's trigger circuit, 81 
Secondary emission, 85, 289 
Seeley and Kimball's linearity measuring 

circuit, 150 
Sensitivity, deflection, 269 

of photographic emulsions, 299 
Series connected amplifier, 219 
Servo simulators, 345 



Shaping circuits, 203, 323, 328, 348 

effect of feed-back amplifier, 207 
Shift potentials, 285 
Simulators, 345 
Single-stroke time base, 41, 52 
Sinusoidal time base, 11, 100 
Skiatron, 298 
Slope impedance, 17 

negative, 70, 356 
Smith's time base, 53 
Speed of photographic emulsions, 299 

of writing, 108, 301 
Spiral time base, 105 
Splitter plate, 262 
Split time base, 228 
Spot size, 283 

stationary, effect of, 53, 266 

velocity, 112, 305 
Square-wave generators, 28, 3 1 , 64, 181, 

3flj 355. 356 , 

generators, multiar, 356 

generators, multivibrator, 28, 356 

generators, transitron, 355 
Squegger, 89 
Squegging oscillator, 89 
Stabilization, 244 
Stabilized frequency divider, 244 

time base, 245 
Stable regime, 52 
Standards, television, 7, 154 
Stationary spot, effect of, 53, 266 
Step functions, 207, 332, 342, 364 
Stray capacitance, 8, 23, 30, 39, 129, 133, 
166, 183, 313, 342, 371 

magnetic fields, 291 
Striking potential, 15, 309 
Strobe valve, 44 
Superposition, principle of, 331 
Suppressor current, negative, 6(i 

grid-controlled sanatron, 181 

locking, 68 

-screen coupling, 55 
Sweep expansion, 41, in, 221 
Switching device, 10, 14 
Synchronization, 23, 34, 39. 232, 2$6 

erratic, 34, 232, 234, 256 

of hard valve time base, 239 

of soft valve time base, 235 

of television receiver, 238 
Synchronizing-pulse generator, 240 

Scantlebury, G. S. P., 256 
Schade, O. H., 97, 2 66 
Scherb, M. V,, 302 
Scherzer, O., 261 
Schetzler, K., 238 
Schmitt, O. H., 53, 8l 
Scroggie, M. G., 223 
Seeley, S. W., 150 
Singer, C, 1 
Smith, L. C, 53 
Smith, J. P., 240 
Somers, F. J., 240 
Sowter, G. A. V., 291 



3 86 

Spencer, R. E., 267 
Spooner, A, M., 279 
Starr, A. T,, 79 
Stevens, W. H., 1 14 
Stoney, G. Johnstone, 259 
Swinton, A, Campbell-, 259 



Taylor's method of linearization, 126 
Television, interlaced scanning, 7, 154 

image, 7, 108, 300 

raster, 7, no, 154 

synchronization, 108, 238, 240 
Three-phase oscillator, 375 
Thyratron, 19, 25, 309 

characteristics, 309 

control ratio, 20, 311 

extinction current, 314 

extinction potential, 19, 22 

striking potential, 19, 311 
Time base, blocking oscillator, 89 

capacitive, 14, 15, 37 

circular, 11, 100 

constant, 32, 36, 72, 117, 176, 185, 356 

direction of rotation, 13 

effective H.T. potential, 130, 132, 

343 
efficiency, 16 
electromagnetic, 95 
expanded, 41, 111,221 
exponential, 14, 16 
externally operated, 52 
high-efficiency, 27, 95, 97 
high-frequency, 181, 186 
inductive, 15, 95 
jitter, 34, 76 
linear saw-tooth, 13 
Miller-capacitance, 158 
modulator, 360 
neon, 15 
performance, 50 
phantastron, 172 
phase opposition, 230 
radial, 106 

repetition frequency, 25 
rotation, direction of, 13 
sanatron, 181 
scalariform, 243, 254 
single-stroke, 42, 52 
soft valve, 15 
spiral, 105 
split, 228 

synchronization, 23, 34, 39, 232, 256 
thyratron, 19, 25, 309 
thyratron characteristics, 309 
thyratron control ratio, 20, 311 
thyratron extinction current, 314 
thyratron extinction potential, 9, 22 
thyratron striking potential, 19, 311 
transitron, 60, 172 
triangular, 36 
triggered, 52 
velocity -modulation, 108 



TIME BASES 



Time, deionization, 309 

discharge, 18 

marker, 8, 143 

modulation, 360 
Tomlinson's method of linearization, 119 
Trace amplitude, 39 

brightness modulation, 273 

direction of rotation, 13 

jitter, 34, 76, 238, 240 

length, 22, 3Q, 43 

linearity, 112, 141 

return black-out, 41, 53 

shift, 148 

synchronization, 34 

velocity, 112, 142, 186 
Transitron, 56 

flip-flop, 64 

Miller time base, 172 

relaxation oscillator, 56 

saw-tooth oscillator, 60 

sinusoidal oscillator, 58 

square -wave generator, 64, 70, 355 
Trant, 64 

cathode-coupled, 70 

screen-coupled, 64 

square-wave generator, 64, 70, 355 
Trapezium, anti-distortion tube, 262, 276 

distortion, 154, 276 
Transfer operators, 210 
Transitron flip-flop, 64 

flip-flop cathode-coupled, 70 

flip-flop screen-coupled, 64 

relaxation oscillator, 56 

saw-tooth oscillator, 60 

sinusoidal oscillator, 58 

square-wave generator, 355 
Trigger circuit, 55 

circuit limiting conditions, 52, 56 

Taylor, M. K., 125 
Thales, r 

Thomson, J. J., 259 
Tihelka, F., 375 
Tolson, W. L., 93 
Tomlinson, T. B., 119, 238 



Unit function, 207, 332, 342, 364 
Unstable regime, 52, 56 
Use of cathode-follower probe, 293 
of cathode ray tube, 289 



INDEX 



3 8 7 



Uttley, A. M., 210 



Valve curvature linearization, 
Velocity error, 112 

of beam, 275 

modulation time base, 108 

modulation television, 108 

spot, 112, 171 

trace, 112, 142, 171 



114 



Virtual earth, 204 
Voltage-doubler, 264 
-multiplier, 266 

Van der Bijl, H. J., 260 

Van der Mark, J., 130 

Van der Pol, B., 56 

Vccchiacchi, F., 32 

Venis, A., 130 

Von Ardenne, M., 17, 104, 142, 192, 261 



Walker's voltage-multiplier, 266 
Watson Watt, Herd and Bainb ridge - 

Bell's time base, 103 
Wave-form generator, in, 363 

-form, non-linearity of, 112 

-form, parabolic, ng, 335 

-form shaping circuits, 203, 323, 328, 

348 

Wave-front distortion, 343 

Wehnelt cylinder, 260 

Wheatcroft effect, 275 

White and Blumlcin's time base, 97 

White and FaudeU's thyratron time base, 
28 

White and FaudeU's time base, 92 

White's blocking oscillator pulse gener- 
ator, 362 

Whiteley's differentiating circuit, 339 

Williams' multiar, 356 

Wilson's method of linearization, 125 



Writing speed, 301 

WaideUch, W. D., 264 

Walker, A. H. B., 266 

Walton, E. T. S., 264 

Watson Watt, R. A., 16, 89 

Wehnelt, A., 259 

Weichert, E., 259 

Weiss, E,, 57 

West, D., 261 

Wheatcroft, E., 275 

White, E. L. C, 28, 32, 60, 91, 92, 93, 

95. 97. 137. 24°. 256, 35&. 362 
Whiteley, J. W., 171, 339. 362 
Williams, B. C. Fleming, 8, 60, 72, 178, 

26a, 277. 356 
Williams, F. C, 32, 83, 106, 114, 135. 

158, 171, 173. «8i, 188, 199, 203, 

210, 250, 323, 341. 345, 346. 357 
Willoner, G., 375 
Wilson, J. C, 125 
Wood, H., 125 



Young, A. J., 60 

Zigzag effect, 317 
Zworykin, V. K., 261